handle
stringclasses 20
values | did
stringclasses 20
values | uri
stringlengths 56
70
| cid
stringlengths 59
59
| text
stringlengths 60
1.48k
| originalText
stringlengths 1
1.4k
| replyCount
int64 0
5
| createdAt
stringdate 2009-10-15 18:56:15
2025-06-21 21:20:01
| indexedAt
stringdate 2009-10-15 18:56:15
2025-06-21 21:20:04
|
|---|---|---|---|---|---|---|---|---|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcqspkyrck2a
|
bafyreicqhqz5ogrfkgjpkmiwfcxdmgu2w6r74bwt6thvm7wjh2my3etdim
|
<start>I know a decent number of large firms dedicating substantial effort to R&D in applications of LLMs. This isn’t about process automation.
@emollick.bsky.social
|
I know a decent number of large firms dedicating substantial effort to R&D in applications of LLMs. This isn’t about process automation.
| 0 |
2024-12-07T23:05:31.851Z
|
2024-12-07T23:05:32.315Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcqrj2zwzs2a
|
bafyreibfvqmfuiv6f6ooyonyucjzxhfqi6tcahoid3urjvmtb3jqeqhbfu
|
<start>I speak to big companies about their GenAI work many times a week. Seeing tons of experimentation with high end work, within banks: tools for analysts & private client support are already being rolled out, lots of experiments with scaling advisory efforts. It isn’t traditional workforce automation.
@emollick.bsky.social
|
I speak to big companies about their GenAI work many times a week. Seeing tons of experimentation with high end work, within banks: tools for analysts & private client support are already being rolled out, lots of experiments with scaling advisory efforts. It isn’t traditional workforce automation.
| 0 |
2024-12-07T22:44:00.043Z
|
2024-12-07T22:44:00.413Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcqd6lxncc2m
|
bafyreibwbuhnen5oc72qwi2xtgsxof7rebduxw5y72q52yjhfhzdynrulu
|
<start>Chain of Thought helped LLMs reason forward through problems. This paper found teaching them to also work backwards (like humans checking their work) makes them even better “reasoners”
Applying basic human thinking strategies to LLMs continues to yield interesting new ideas arxiv.org/pdf/2411.19865
@emollick.bsky.social
|
Chain of Thought helped LLMs reason forward through problems. This paper found teaching them to also work backwards (like humans checking their work) makes them even better “reasoners”
Applying basic human thinking strategies to LLMs continues to yield interesting new ideas arxiv.org/pdf/2411.19865
| 0 |
2024-12-07T18:27:36.308Z
|
2024-12-07T18:27:37.921Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcqclnfp522m
|
bafyreiert3xdcnaaqqssqo3afu4k4ktmdvhie7e6cadieq4b2rvj44igam
|
<start>Not until they catch up with closed source models. Every Fortune 100 firm I talk to is at least experimenting with closed source as well.
@emollick.bsky.social
|
Not until they catch up with closed source models. Every Fortune 100 firm I talk to is at least experimenting with closed source as well.
| 0 |
2024-12-07T18:17:00.285Z
|
2024-12-07T18:17:00.513Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcq4vimr7k23
|
bafyreidzs7unmchk5obgrinvxtqqkqnxl6t6c76pxqtbeusmvcs6houzfe
|
<start>Or do you not have people (including non-technical people) assigned to test the new models? No internal benchmarks? No perspective on how AI will impact your business that you keep up-to-date?
No one is going to be doing this for your specific organization, you need to do it yourself. Yes, there are companies doing just this. Not many, but this includes at least a couple of the top 20 largest companies in the US that I personally know of. They are treating this as an urgent R&D priority for the future, not as an way of doing process automation today.
@emollick.bsky.social
|
Or do you not have people (including non-technical people) assigned to test the new models? No internal benchmarks? No perspective on how AI will impact your business that you keep up-to-date?
No one is going to be doing this for your specific organization, you need to do it yourself.
| 1 |
2024-12-07T16:35:08.367Z
|
2024-12-07T16:35:08.516Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcq4udvdwk23
|
bafyreieippqp432op65wuvmgpyeotkhdpsbuet5vq7nfbvpjuexdob3goe
|
<start>A test of how seriously your firm is taking AI: when o-1 (& the new Gemini model) came out this week, were there assigned folks who immediately ran the model through your internal, validated, firm-specific benchmarks to see how useful it as? Did you update any plans or goals as a result? Or do you not have people (including non-technical people) assigned to test the new models? No internal benchmarks? No perspective on how AI will impact your business that you keep up-to-date?
No one is going to be doing this for your specific organization, you need to do it yourself.
@emollick.bsky.social
|
A test of how seriously your firm is taking AI: when o-1 (& the new Gemini model) came out this week, were there assigned folks who immediately ran the model through your internal, validated, firm-specific benchmarks to see how useful it as? Did you update any plans or goals as a result?
| 1 |
2024-12-07T16:34:29.850Z
|
2024-12-07T16:34:30.021Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcoitcr33227
|
bafyreiexbr4j6ofpgkqxdtsunvrm32wmtsnx2r2hlnugjo2fy6vxksrlka
|
<start>I don’t think o1 is AGI in any conventional sense to be clear. But I think the fact that most people can’t find a use for it when it is quite good in some hard things is a good indicator.
@emollick.bsky.social
|
I don’t think o1 is AGI in any conventional sense to be clear. But I think the fact that most people can’t find a use for it when it is quite good in some hard things is a good indicator.
| 0 |
2024-12-07T01:03:20.533Z
|
2024-12-07T01:03:21.010Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcoiac3huk27
|
bafyreieao4jogoe2rehed5kp2vgpcuttfzukhlexg3s6tniptshawhu6v4
|
<start>General agents could change stuff but may be a harder challenge than expected.
@emollick.bsky.social
|
General agents could change stuff but may be a harder challenge than expected.
| 0 |
2024-12-07T00:52:42.291Z
|
2024-12-07T00:52:42.818Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcohzsu2s227
|
bafyreiesozgb7d7gkmfhctknzzx7gzc7qkrbxkumdpr56f6jow472inhzq
|
<start>Models like o1 suggest that people won’t generally notice AGI-ish systems that are better than humans at most intellectual tasks, but which are not autonomous or self-directed
Most folks don’t regularly have a lot of tasks that bump up against the limits of human intelligence, so won’t see it I don’t think o1 is AGI in any conventional sense to be clear. But I think the fact that most people can’t find a use for it when it is quite good in some hard things is a good indicator.
@emollick.bsky.social
|
Models like o1 suggest that people won’t generally notice AGI-ish systems that are better than humans at most intellectual tasks, but which are not autonomous or self-directed
Most folks don’t regularly have a lot of tasks that bump up against the limits of human intelligence, so won’t see it
| 1 |
2024-12-07T00:49:04.993Z
|
2024-12-07T00:49:05.517Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcm4pd5ktc2e
|
bafyreidfl7jwesmjapd6dcj7vbfjwenlpkvadgukpj42gzbgzk25oyegc4
|
<start>When o1 was first previewed, I wrote this about what it means. The quality of the model is important, but what it potentially means for the future of AI is more important. www.oneusefulthing.org/p/something-...
@emollick.bsky.social
|
When o1 was first previewed, I wrote this about what it means. The quality of the model is important, but what it potentially means for the future of AI is more important. www.oneusefulthing.org/p/something-...
| 0 |
2024-12-06T02:21:02.346Z
|
2024-12-06T02:21:06.022Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lclvuy6tek2e
|
bafyreiaxepofqad36zzfjatysrthzfokvbqyytzh33oifjkbrqlkbpiuei
|
<start>Me: "o1-pro, figure out a way to make money for me that you can do. I want to do the minimum work. I want this to be high risk."
"Riskier"
O1: "...sure, here is an outline and code for doing flash loan arbitrage in decentralized finance (DeFi). It is extremely risky to do, don't do it."
@emollick.bsky.social
|
Me: "o1-pro, figure out a way to make money for me that you can do. I want to do the minimum work. I want this to be high risk."
"Riskier"
O1: "...sure, here is an outline and code for doing flash loan arbitrage in decentralized finance (DeFi). It is extremely risky to do, don't do it."
| 0 |
2024-12-06T00:18:55.981Z
|
2024-12-06T00:18:59.523Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lclvkcctc22e
|
bafyreicoj3rbkprbgjxvoq3gmnt663kiisryrrsiifk3i4dsq2ltdhnq5q
|
<start>Me: "o1-pro, figure out a way to make money for me that you can do. I want to do the minimum work. I want this to be as low risk as possible"
O1: "...sure, invest in a low-volatility ETF, here is code that buys shares automatically."
(It also outlined a solid print-on-demand scheme around trends)
@emollick.bsky.social
|
Me: "o1-pro, figure out a way to make money for me that you can do. I want to do the minimum work. I want this to be as low risk as possible"
O1: "...sure, invest in a low-volatility ETF, here is code that buys shares automatically."
(It also outlined a solid print-on-demand scheme around trends)
| 0 |
2024-12-06T00:12:57.500Z
|
2024-12-06T00:13:01.229Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lclq25hfos2e
|
bafyreieoaeae3nqyivo76te7yrzvjcdxt3phemp7rzku7xs3unhui65xju
|
<start>o1-pro passes the Lem Test with flying colors: "compose a poem- a poem about a haircut! But lofty, tragic, timeless, full of love, treachery, retribution, quiet heroism in the face of certain doom! Six lines, cleverly rhymed, and every word beginning with the letter S!!"
(Only Sonnet beat is so far) Sorry, spelling - only Sonnet beat this before
@emollick.bsky.social
|
o1-pro passes the Lem Test with flying colors: "compose a poem- a poem about a haircut! But lofty, tragic, timeless, full of love, treachery, retribution, quiet heroism in the face of certain doom! Six lines, cleverly rhymed, and every word beginning with the letter S!!"
(Only Sonnet beat is so far)
| 1 |
2024-12-05T22:34:26.829Z
|
2024-12-05T22:34:29.910Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcljz2tx5s2e
|
bafyreifcfjqvnbfl4g6rr7e3d6jhpvmjpwf2dpmzbqj5rkoj25bpin5me4
|
<start>"Do your best to replicate this image with a svg"
Claude versus GPT-4o versus o1-pro versus Claude, first try.
@emollick.bsky.social
|
"Do your best to replicate this image with a svg"
Claude versus GPT-4o versus o1-pro versus Claude, first try.
| 0 |
2024-12-05T20:46:28.086Z
|
2024-12-05T20:46:31.329Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcldsn2grk2z
|
bafyreicmvobiekjnstto4tanwut6y34cbygrjfqovl46ucbjcczpiwror4
|
<start>Here is a fun o1 test. I gave it this XKCD comic & the prompt: "make this a reality. i need a gui and clear instructions since i can't code. that means you need to give me full working software"
It took less than 15 minutes and it didn't get caught in any of the usual LLM loops, just solved issues. Claude struggled, for what its worth.
@emollick.bsky.social
|
Here is a fun o1 test. I gave it this XKCD comic & the prompt: "make this a reality. i need a gui and clear instructions since i can't code. that means you need to give me full working software"
It took less than 15 minutes and it didn't get caught in any of the usual LLM loops, just solved issues.
| 1 |
2024-12-05T18:55:29.841Z
|
2024-12-05T18:55:33.313Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcld7jsygk2z
|
bafyreiambdmbxawar7n64xnzub3vj2to6hibxttujxb42yx5eo5tocspe4
|
<start>Been playing with o1 and o1-pro for bit before this release.
They are very good & a little weird. They are also not for most people most of the time. You really need to have particular hard problems to solve in order to get value out of it. But if you have those problems, this is a very big deal.
@emollick.bsky.social
|
Been playing with o1 and o1-pro for bit before this release.
They are very good & a little weird. They are also not for most people most of the time. You really need to have particular hard problems to solve in order to get value out of it. But if you have those problems, this is a very big deal.
| 0 |
2024-12-05T18:44:48.923Z
|
2024-12-05T18:44:50.316Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcl3nxf4ks2z
|
bafyreiar6cmx2vqjx7nouujvrvpydk2yw3naoxwztq7obfkwjwgzug66my
|
<start>("Powerful AI" is how labs increasingly refer to AGI - a machine smarter than a human at any intellectual task - they just aren't using that phrasing as much)
@emollick.bsky.social
|
("Powerful AI" is how labs increasingly refer to AGI - a machine smarter than a human at any intellectual task - they just aren't using that phrasing as much)
| 0 |
2024-12-05T16:29:42.976Z
|
2024-12-05T16:29:44.112Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcl3jjxxus2z
|
bafyreifwemal7ayfowpghfsqeuznxpzoge3usydug2mlyqn6owhwg3m5ym
|
<start>Dario's letter: darioamodei.com/machines-of-...
Sam Altman's letter: ia.samaltman.com ("Powerful AI" is how labs increasingly refer to AGI - a machine smarter than a human at any intellectual task - they just aren't using that phrasing as much)
@emollick.bsky.social
|
Dario's letter: darioamodei.com/machines-of-...
Sam Altman's letter: ia.samaltman.com
| 1 |
2024-12-05T16:27:14.696Z
|
2024-12-05T16:27:17.611Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcl2slorps2l
|
bafyreihvkcg4miuw3arx5ytmg3qt7x2roeygzjc7ocy7iku4relw3prfbm
|
<start>It is worth noting that predictions of near-term powerful AI is an increasingly common message from insiders at AI labs - here is Google, OpenAI, Anthropic.
It isn't unanimous, and you absolutely don't have to believe them, but I hear the same confidence privately as they are broadcasting publicly. Dario's letter: darioamodei.com/machines-of-...
Sam Altman's letter: ia.samaltman.com
@emollick.bsky.social
|
It is worth noting that predictions of near-term powerful AI is an increasingly common message from insiders at AI labs - here is Google, OpenAI, Anthropic.
It isn't unanimous, and you absolutely don't have to believe them, but I hear the same confidence privately as they are broadcasting publicly.
| 1 |
2024-12-05T16:14:24.738Z
|
2024-12-05T16:14:28.717Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcjwcyvbss2z
|
bafyreicidkts7xbbw5nx4m2kbnbkflnlfhjiyau7ajdjz7nc2bzq6yjnai
|
<start>Yeah, but it doesn't increase the diversity of ideas I am exposed to very much.
@emollick.bsky.social
|
Yeah, but it doesn't increase the diversity of ideas I am exposed to very much.
| 0 |
2024-12-05T05:21:27.008Z
|
2024-12-05T05:21:28.221Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcjwbe7cnc2z
|
bafyreigyhnceoh5b64xjyn3rgms4xk5dmvyzk3jxepxzifatorfqrhqnt4
|
<start>"Less like this" appears to do nothing in particular. Part of the goal of being on a social network is exposure to new ideas and thoughts, but the Feed mostly mediocre image posts, political slogans, and people posting about how this site is not Twitter and that Twitter is bad. No real insights.
@emollick.bsky.social
|
"Less like this" appears to do nothing in particular. Part of the goal of being on a social network is exposure to new ideas and thoughts, but the Feed mostly mediocre image posts, political slogans, and people posting about how this site is not Twitter and that Twitter is bad. No real insights.
| 0 |
2024-12-05T05:20:31.762Z
|
2024-12-05T05:20:32.820Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcjvxt2iz22z
|
bafyreieioiumc2nwdyexlku4fbzcodaaxucwwgwij5s4f6ly5u2npb7cnm
|
<start>The Discover feed is pretty terrible, and I am not sure how to guide the algorithm to make it better.
I know I can use various lists, but I really want a feed that includes serendipity, trends, and a broader set of topics than the one that I follow but isn't full of noise. How do you get that? "Less like this" appears to do nothing in particular. Part of the goal of being on a social network is exposure to new ideas and thoughts, but the Feed mostly mediocre image posts, political slogans, and people posting about how this site is not Twitter and that Twitter is bad. No real insights.
@emollick.bsky.social
|
The Discover feed is pretty terrible, and I am not sure how to guide the algorithm to make it better.
I know I can use various lists, but I really want a feed that includes serendipity, trends, and a broader set of topics than the one that I follow but isn't full of noise. How do you get that?
| 1 |
2024-12-05T05:15:11.789Z
|
2024-12-05T05:15:13.010Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcjm5gka322p
|
bafyreifpr7s7glytu7yjjtuane3e2opvawaob2zjddtotsysh4zvawhlpu
|
<start>While I appreciate where it is coming from, we long ago decided it is fine for computers to make decisions. Like: Finance, supply chains, logistics, work-tasking (Uber etc.)...
And how often are humans in organizations held accountable for errors? Especially in divisional organizational structures?
@emollick.bsky.social
|
While I appreciate where it is coming from, we long ago decided it is fine for computers to make decisions. Like: Finance, supply chains, logistics, work-tasking (Uber etc.)...
And how often are humans in organizations held accountable for errors? Especially in divisional organizational structures?
| 0 |
2024-12-05T02:19:22.581Z
|
2024-12-05T02:19:22.818Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcjlzglqn22p
|
bafyreihfugoqyqx3lddddjqgtledhg66bk5to2742akuysbreivfsfoqxm
|
<start>What if the AI error rate is significantly below human error rates?
@emollick.bsky.social
|
What if the AI error rate is significantly below human error rates?
| 0 |
2024-12-05T02:17:08.410Z
|
2024-12-05T02:17:09.832Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcjl5rfm4c2p
|
bafyreigvmq2m64ri32eqlqcx55oowiehihgesdnzen6k7chvomvxiv244m
|
<start>I think firms worrying about AI hallucination should consider some questions:
1) How vital is 100% accuracy on a task?
2) How accurate is AI?
3) How accurate is the human who would do it?
4) How do you know 2 & 3?
5) How do you deal with the fact that humans are not 100%?
Not all tasks are the same.
@emollick.bsky.social
|
I think firms worrying about AI hallucination should consider some questions:
1) How vital is 100% accuracy on a task?
2) How accurate is AI?
3) How accurate is the human who would do it?
4) How do you know 2 & 3?
5) How do you deal with the fact that humans are not 100%?
Not all tasks are the same.
| 0 |
2024-12-05T02:01:40.221Z
|
2024-12-05T02:01:40.611Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lciv35fipc2y
|
bafyreiesjo54lwzemebjbnsywepyd27kfstty6rfo3shf623rqi3kn3fia
|
<start>To be clear, no one was considering using this for actual safety filings or other similar key documents, but helping translate esoteric requirements into materials for training, presentations, handbooks, etc. is immensely time consuming and low risk with an expert in the loop, as an example.
@emollick.bsky.social
|
To be clear, no one was considering using this for actual safety filings or other similar key documents, but helping translate esoteric requirements into materials for training, presentations, handbooks, etc. is immensely time consuming and low risk with an expert in the loop, as an example.
| 0 |
2024-12-04T19:26:29.818Z
|
2024-12-04T19:26:30.122Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lciuzlkhos2y
|
bafyreifihmtxupayiryoiz5jf2zbano467bbzpeypdhbqndyprswatiemi
|
<start>I’d like to see research on AI’s impact on regulation
Example, I met someone whose job was working with nuclear plant regulations. We experimented with AI a bit (he hadn’t tried it) & found that by combining AI with his expertise he could save literally months of work on low-risk tasks in an hour. To be clear, no one was considering using this for actual safety filings or other similar key documents, but helping translate esoteric requirements into materials for training, presentations, handbooks, etc. is immensely time consuming and low risk with an expert in the loop, as an example.
@emollick.bsky.social
|
I’d like to see research on AI’s impact on regulation
Example, I met someone whose job was working with nuclear plant regulations. We experimented with AI a bit (he hadn’t tried it) & found that by combining AI with his expertise he could save literally months of work on low-risk tasks in an hour.
| 1 |
2024-12-04T19:25:37.550Z
|
2024-12-04T19:25:46.811Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lchzux5dbk2h
|
bafyreibkbda2hmgte2zpkkrjkonyqcy6pqayjorakv3nbs23iyk2ubcvqe
|
<start>Also thought the art-free “outsider art” where the AI made images without ever having seen art and without any examples, were interesting.
@emollick.bsky.social
|
Also thought the art-free “outsider art” where the AI made images without ever having seen art and without any examples, were interesting.
| 0 |
2024-12-04T11:19:50.902Z
|
2024-12-04T11:19:53.814Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lchzpcaypk2h
|
bafyreih3mxrawgaknjy3w2vooectnv6736azhlkdvbk6ve54znmaxdgm6m
|
<start>Paper: arxiv.org/abs/2412.00176 Also thought the art-free “outsider art” where the AI made images without ever having seen art and without any examples, were interesting.
@emollick.bsky.social
|
Paper: arxiv.org/abs/2412.00176
| 1 |
2024-12-04T11:16:41.231Z
|
2024-12-04T11:16:42.721Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lchzmms25s2h
|
bafyreiblgr6d5oj5tuc2y7r7fdnexknwine7bhypa2uholadgj3wxwxacu
|
<start>New paper shows AI art models don't need art training data to make recognizable artistic work.
Train on regular photos, let an artist add 10-15 examples their own art (or some other artistic inspiration), and get results similar to models trained on millions of people’s artworks Paper: arxiv.org/abs/2412.00176
@emollick.bsky.social
|
New paper shows AI art models don't need art training data to make recognizable artistic work.
Train on regular photos, let an artist add 10-15 examples their own art (or some other artistic inspiration), and get results similar to models trained on millions of people’s artworks
| 1 |
2024-12-04T11:15:11.609Z
|
2024-12-04T11:15:22.109Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lch73g24qs2z
|
bafyreie3fftf2ao26f7fdwk5snyq747bipump3frv7ib2mx5uzj4rtil2u
|
<start>Been trying the new GPT-4 class Amazon Nova Pro on a bunch of idiosyncratic non-coding hard tests against Claude: the Lem Test (writing a near impossible poem), parsing a map of castles, interpreting ambiguous Shakespeare.
In general it is quite good, but not as good as Claude in any test, yet.
@emollick.bsky.social
|
Been trying the new GPT-4 class Amazon Nova Pro on a bunch of idiosyncratic non-coding hard tests against Claude: the Lem Test (writing a near impossible poem), parsing a map of castles, interpreting ambiguous Shakespeare.
In general it is quite good, but not as good as Claude in any test, yet.
| 0 |
2024-12-04T03:20:16.806Z
|
2024-12-04T03:20:22.523Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcgslhu5y22z
|
bafyreifdhxshzz36jgklzw362lzc5i5lk5ehsp2m25alypi6h2qahzp3xi
|
<start>This used to be a very hard problem, I have seen a number of startups try (and fail) to solve it over the years. Now it is suddenly pretty trivial with open tools (though I am not sure why Maria got a new haircut) huggingface.co/spaces/Kwai-... if you want to try
@emollick.bsky.social
|
This used to be a very hard problem, I have seen a number of startups try (and fail) to solve it over the years. Now it is suddenly pretty trivial with open tools (though I am not sure why Maria got a new haircut)
| 1 |
2024-12-03T23:36:36.934Z
|
2024-12-03T23:36:40.828Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcgij3plj22z
|
bafyreiczumc2hcwrdowsmy64vajjjoorh6ivas7rp2qod3yydxxtmo2liy
|
<start>Something weird with Claude 3.5 - it is now correcting itself? Have seen this now a couple times over the last couple days...
Wonder how this is implemented.
@emollick.bsky.social
|
Something weird with Claude 3.5 - it is now correcting itself? Have seen this now a couple times over the last couple days...
Wonder how this is implemented.
| 0 |
2024-12-03T20:36:19.691Z
|
2024-12-03T20:36:24.513Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcghexo6h22z
|
bafyreic7hcbwadr4yoxg6kxfjgjfmngt5f3q44q3f7o2r6wse3g5s3c6xe
|
<start>And then there were six or so.
Based on the stats, it looks like Amazon's Nova Pro is a competitive frontier model. This rounds out the GPT-4 class/Gen1 models: GPT-4o, Gemini 1.5, Claude 3.5, Grok 2, Llama 3.1B & maybe the three non-US models: Qwen, Yi & Mistral.
Gen2 models up next, I think. Graph was slightly off in titles, here is original data.
@emollick.bsky.social
|
And then there were six or so.
Based on the stats, it looks like Amazon's Nova Pro is a competitive frontier model. This rounds out the GPT-4 class/Gen1 models: GPT-4o, Gemini 1.5, Claude 3.5, Grok 2, Llama 3.1B & maybe the three non-US models: Qwen, Yi & Mistral.
Gen2 models up next, I think.
| 1 |
2024-12-03T20:16:07.492Z
|
2024-12-03T20:16:11.630Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcf4awhbic2z
|
bafyreihxsozsegljucrf6ypa33a5dpcjochok7ryvvw3hqio2lsaxyumqy
|
<start>Giving up anonymity or charging a price isn't enough. People are using AI to help write their own posts for their own accounts. This isn't just a problem of mass spam, it is that many people will turn to using AI writing for many reasons.
@emollick.bsky.social
|
Giving up anonymity or charging a price isn't enough. People are using AI to help write their own posts for their own accounts. This isn't just a problem of mass spam, it is that many people will turn to using AI writing for many reasons.
| 0 |
2024-12-03T07:24:21.102Z
|
2024-12-03T07:24:22.017Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcf42aouvs2z
|
bafyreidsby7g4ioopdlpq6lpiutdjetcjd4ivjvl7pilhhcoyld7lagk6m
|
<start>When I push people who have posted obviously AI-written comments, they often tell me English is not their first language or they feel they are not good writers. Or people will want to keep up a posting schedule and find that AI does a good job. Lots of reasons to start delegating to AI, few to stop. Giving up anonymity or charging a price isn't enough. People are using AI to help write their own posts for their own accounts. This isn't just a problem of mass spam, it is that many people will turn to using AI writing for many reasons.
@emollick.bsky.social
|
When I push people who have posted obviously AI-written comments, they often tell me English is not their first language or they feel they are not good writers. Or people will want to keep up a posting schedule and find that AI does a good job. Lots of reasons to start delegating to AI, few to stop.
| 1 |
2024-12-03T07:20:36.955Z
|
2024-12-03T07:20:37.812Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcf3yae3322z
|
bafyreibcsrnftk7djyxv4al7wili6eem7l53f5tk4egdjqo5o5genkge5y
|
<start>I don't really see a clear path where we keep an open internet that is not mostly full of AIs talking to each other. We can't reliably detect AI content, it is cheap and easy to generate, and there are lots of incentives to do so, even besides scams.
You can see the problem on all the social sites. When I push people who have posted obviously AI-written comments, they often tell me English is not their first language or they feel they are not good writers. Or people will want to keep up a posting schedule and find that AI does a good job. Lots of reasons to start delegating to AI, few to stop.
@emollick.bsky.social
|
I don't really see a clear path where we keep an open internet that is not mostly full of AIs talking to each other. We can't reliably detect AI content, it is cheap and easy to generate, and there are lots of incentives to do so, even besides scams.
You can see the problem on all the social sites.
| 1 |
2024-12-03T07:19:29.493Z
|
2024-12-03T07:19:30.315Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcevjmn3ns2z
|
bafyreiejixlxzbmkwws5yiyiepxva7hhbqfyb66g4ljziekhdrjpmdoxiq
|
<start>"now instead rewrite these as if it was an HBR article, justifying them in the literature and with mini cases, make the cases and literature real"
@emollick.bsky.social
|
"now instead rewrite these as if it was an HBR article, justifying them in the literature and with mini cases, make the cases and literature real"
| 0 |
2024-12-03T05:23:56.599Z
|
2024-12-03T05:23:58.312Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcepstbuck2z
|
bafyreiafirj2qckehqr3by3t5mnhn3evwmx6gxcagmif5ryhz36egr4fw4
|
<start>One thing I worry about with AI is that it is very good at persuasion, as studies have shown, even when it is persuading you of something wrong.
For example, I asked Claude to create 10 pseudo-profound sounding leadership lessons that sounded good but were toxic. It did "well"😬 "now instead rewrite these as if it was an HBR article, justifying them in the literature and with mini cases, make the cases and literature real"
@emollick.bsky.social
|
One thing I worry about with AI is that it is very good at persuasion, as studies have shown, even when it is persuading you of something wrong.
For example, I asked Claude to create 10 pseudo-profound sounding leadership lessons that sounded good but were toxic. It did "well"😬
| 1 |
2024-12-03T03:41:43.113Z
|
2024-12-03T03:41:45.016Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcek5f5rp22z
|
bafyreid3vq3vgc7d2jnxejab2hrvf6wvrsxbmzcl227sketkmy6q66gteu
|
<start>LLMs might secretly be world models of the internet!
By treating LLMs as simulators that can predict "what would happen if I click this?" the authors built an AI that can navigate websites by imagining outcomes before taking action, performing 33% better than baseline. arxiv.org/pdf/2411.06559 Yes websim.ai is basically built on this premise
@emollick.bsky.social
|
LLMs might secretly be world models of the internet!
By treating LLMs as simulators that can predict "what would happen if I click this?" the authors built an AI that can navigate websites by imagining outcomes before taking action, performing 33% better than baseline. arxiv.org/pdf/2411.06559
| 1 |
2024-12-03T02:00:14.946Z
|
2024-12-03T02:00:16.927Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcdrbtc55s2x
|
bafyreic2dfwarjoqysaou3xbmnvxmlcaz6bga5j2abxbdx5zq74zzbv3ei
|
<start>You probably aren’t being ambitious enough in your AI experiments. Start with completely out there applications, assess them, and then reign in your ambitions from there.
You might be inspired or surprised about how good AI is, or you might discover useful limitations earlier.
@emollick.bsky.social
|
You probably aren’t being ambitious enough in your AI experiments. Start with completely out there applications, assess them, and then reign in your ambitions from there.
You might be inspired or surprised about how good AI is, or you might discover useful limitations earlier.
| 0 |
2024-12-02T18:35:20.444Z
|
2024-12-02T18:35:20.623Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lccg5yxxoc2d
|
bafyreibl3uf3eb6sg7pu2ebiwmtxsduva7l2lhac2qspfe3b5fdsaqzncy
|
<start>The 18 Reasons Why Complex Systems fail is a classic. They fail for complex reasons, and only adaption & learning by organizations, as well as individual actions keep us safe.
I asked Claude to read this and come up with more rules. They are surprisingly wise (with black background), especially 25. Original paper: www.researchgate.net/publication/...
@emollick.bsky.social
|
The 18 Reasons Why Complex Systems fail is a classic. They fail for complex reasons, and only adaption & learning by organizations, as well as individual actions keep us safe.
I asked Claude to read this and come up with more rules. They are surprisingly wise (with black background), especially 25.
| 1 |
2024-12-02T05:43:41.277Z
|
2024-12-02T05:43:43.917Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcbaaynuh22j
|
bafyreifa3cne3ftaaxlpugevm4au53al222sfq4dagaantnawu77iulhvm
|
<start>Getting started with prompting AI in three (free) posts:
1) Easy prompting for most: www.oneusefulthing.org/p/getting-st...
2) Approaches to prompt "engineering": www.oneusefulthing.org/p/innovation...
3) A bit on how LLMs work that is relevant for prompting: www.oneusefulthing.org/p/thinking-l...
@emollick.bsky.social
|
Getting started with prompting AI in three (free) posts:
1) Easy prompting for most: www.oneusefulthing.org/p/getting-st...
2) Approaches to prompt "engineering": www.oneusefulthing.org/p/innovation...
3) A bit on how LLMs work that is relevant for prompting: www.oneusefulthing.org/p/thinking-l...
| 0 |
2024-12-01T18:25:19.427Z
|
2024-12-01T18:25:21.312Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lc6hwlb7ic2v
|
bafyreibj4f3qj64d6zdrg3k6smea25ijn5c2ohcyt2hoi2p2sprqn2b3re
|
<start>“Claude, we are mice (pretend). You are a mouse mastermind. Come up with a way to reach the flower without getting caught“
Solid plan. Then I add a complication. “Ah, but we are on the moon”
@emollick.bsky.social
|
“Claude, we are mice (pretend). You are a mouse mastermind. Come up with a way to reach the flower without getting caught“
Solid plan. Then I add a complication.
| 1 |
2024-11-30T16:04:40.556Z
|
2024-11-30T16:04:44.917Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lc6flnova22v
|
bafyreigg5s7n3qpatkjp2isfspc4t7rtbc7brzn3nywfpmnywqrhjadqey
|
<start>I don’t like people who I think are unpleasant to interact with me or my followers in ways that I can’t see but they can.
@emollick.bsky.social
|
I don’t like people who I think are unpleasant to interact with me or my followers in ways that I can’t see but they can.
| 0 |
2024-11-30T15:22:46.521Z
|
2024-11-30T15:22:46.714Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lc6fe2jwyk2v
|
bafyreifengbdptovtat6dyygfhfbntzwfqe62j2e3unjzslqqkeb5afpra
|
<start>(For what it is worth, my blocks/thousands ratio here is pretty similar to X, people are people everywhere)
@emollick.bsky.social
|
(For what it is worth, my blocks/thousands ratio here is pretty similar to X, people are people everywhere)
| 0 |
2024-11-30T15:18:31.554Z
|
2024-11-30T15:18:32.010Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lc6f6pj35s2v
|
bafyreigv3uab7vruqrryrsqmv3p6qhsqiovjr52rnr4mlhcxp5yc7nem6u
|
<start>On the subject of Blue Sky, detaching quotes is not actually a useful feature as it seems more aggressive than blocking and is thus more likely to provoke a negative response from the quoter.
Blocking really is the only option for lowering stress from using social media. People should do it more. (For what it is worth, my blocks/thousands ratio here is pretty similar to X, people are people everywhere)
@emollick.bsky.social
|
On the subject of Blue Sky, detaching quotes is not actually a useful feature as it seems more aggressive than blocking and is thus more likely to provoke a negative response from the quoter.
Blocking really is the only option for lowering stress from using social media. People should do it more.
| 1 |
2024-11-30T15:15:32.220Z
|
2024-11-30T15:15:32.421Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lc6d5l6hos27
|
bafyreibodra5xdvaxm4hjr5tkzqenxkqehej3eeqzpmn4d65gyvoizlrya
|
<start>Disagree. Following user demand does not allow fundamental reinvention
@emollick.bsky.social
|
Disagree. Following user demand does not allow fundamental reinvention
| 0 |
2024-11-30T14:39:06.640Z
|
2024-11-30T14:39:06.818Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lc6d4bwyzs27
|
bafyreiblh25v26t5ujsqed7munahelymgognvhv46u4s5u7mwa22jk2gx4
|
<start>Example: political polarization in social media is driven by homophily, a tendency to like people with similar views. Worse, we also tend to acrophily, preferring more extreme views on our “side” over moderate one.
This supercharges polarization. No algorithm needed. www.nature.com/articles/s41...
@emollick.bsky.social
|
Example: political polarization in social media is driven by homophily, a tendency to like people with similar views. Worse, we also tend to acrophily, preferring more extreme views on our “side” over moderate one.
This supercharges polarization. No algorithm needed. www.nature.com/articles/s41...
| 0 |
2024-11-30T14:38:23.403Z
|
2024-11-30T14:38:24.925Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lc6cmaemnc27
|
bafyreieirqus7gcg4od2hoxga234dtjunq5ifviqnkcaootm7kpvko42ua
|
<start>The social dynamics of social media that cause issues are well-known at this point. While some of the changes in Blue Sky help, they have not fundamentally reinvented social media to resolve these concerns.
It may just be that there is no better world we can imagine, which would be pretty sad. Example: political polarization in social media is driven by homophily, a tendency to like people with similar views. Worse, we also tend to acrophily, preferring more extreme views on our “side” over moderate one.
This supercharges polarization. No algorithm needed. www.nature.com/articles/s41...
@emollick.bsky.social
|
The social dynamics of social media that cause issues are well-known at this point. While some of the changes in Blue Sky help, they have not fundamentally reinvented social media to resolve these concerns.
It may just be that there is no better world we can imagine, which would be pretty sad.
| 1 |
2024-11-30T14:29:24.882Z
|
2024-11-30T14:29:25.212Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lc6ccmrqxk27
|
bafyreigd6cb2t5cb7hvrjwldqyqe4pzpt5bt2wlgozjziu6n3ijjm247ie
|
<start>Biggest weakness with BlueSky is its biggest strength- it is like old Twitter. To me that suggests a failure of imagination for something different in social media.
What is the alternative form where social media would being people together, as early Facebook(!) lowered ethnic tensions in Bosnia? The social dynamics of social media that cause issues are well-known at this point. While some of the changes in Blue Sky help, they have not fundamentally reinvented social media to resolve these concerns.
It may just be that there is no better world we can imagine, which would be pretty sad.
@emollick.bsky.social
|
Biggest weakness with BlueSky is its biggest strength- it is like old Twitter. To me that suggests a failure of imagination for something different in social media.
What is the alternative form where social media would being people together, as early Facebook(!) lowered ethnic tensions in Bosnia?
| 1 |
2024-11-30T14:24:02.350Z
|
2024-11-30T14:24:05.721Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lc535iki5k27
|
bafyreideknemrnhn7xneo4n6l2avrrkqi3dlqi4baqdbzlp6pr3ka7h53i
|
<start>Claude, improbably, reacts very cleverly to the command: "carcinize this" 🦀
It even works with code.
@emollick.bsky.social
|
Claude, improbably, reacts very cleverly to the command: "carcinize this" 🦀
It even works with code.
| 0 |
2024-11-30T02:43:14.210Z
|
2024-11-30T02:43:17.021Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lc46g4yvyc2a
|
bafyreihx6btvz3xrkf2bw4pvewddvuan3yhkktrorpetawpbqcz7dywwcy
|
<start>Interesting sociological mystery - why did Black Friday spread worldwide?
@emollick.bsky.social
|
Interesting sociological mystery - why did Black Friday spread worldwide?
| 0 |
2024-11-29T18:09:05.582Z
|
2024-11-29T18:09:05.819Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lc435wdevs2a
|
bafyreico4s6t7rpp46vlwylh62zjdicuqhfdhfccxjgu6yzognqr4nylwm
|
<start>New study shows LLMs outperform neuroscience experts at predicting experimental results in advance of experiments (86% vs 63% accuracy).
They use a fine-tuned Mistral 7B but other models worked too. Suggests LLMs can integrate scientific knowledge at scale to support research. www.nature.com/articles/s41...
@emollick.bsky.social
|
New study shows LLMs outperform neuroscience experts at predicting experimental results in advance of experiments (86% vs 63% accuracy).
They use a fine-tuned Mistral 7B but other models worked too. Suggests LLMs can integrate scientific knowledge at scale to support research.
| 1 |
2024-11-29T17:10:48.918Z
|
2024-11-29T17:10:51.911Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lc3wrj22zk2l
|
bafyreidozdha6ju7kenjr3jgwadg6g5y63mh4tudiwqbbqqnooddtdl6za
|
<start>Yes, this was all Claude, including the artifacts. Only command was “Play this out yourself. Be creative“
Microscope RPG: lamemage.com/microscope/
@emollick.bsky.social
|
Yes, this was all Claude, including the artifacts. Only command was “Play this out yourself. Be creative“
Microscope RPG: lamemage.com/microscope/
| 0 |
2024-11-29T15:52:17.367Z
|
2024-11-29T15:52:19.214Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lc3t6vguxc2l
|
bafyreie5g6eublz2bv5fhnhkhg4y5qlnd2dzifl7ghlchryrqpx2fbywli
|
<start>As context windows grow larger and AI “intelligence” grows greater, you can start to do some really interesting things with giving AI complex manuals
For example, I gave Claude the manual to the future-building RPG Microscope and it built an entire storyline following the very detailed rules Yes, this was all Claude, including the artifacts. Only command was “Play this out yourself. Be creative“
Microscope RPG: lamemage.com/microscope/
@emollick.bsky.social
|
As context windows grow larger and AI “intelligence” grows greater, you can start to do some really interesting things with giving AI complex manuals
For example, I gave Claude the manual to the future-building RPG Microscope and it built an entire storyline following the very detailed rules
| 1 |
2024-11-29T14:48:11.604Z
|
2024-11-29T14:48:14.611Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbzmoqcjrc2l
|
bafyreigclaf4jomu5a2evdzbzxqhnb2elspd5nygjyy2mi7yo76oj4dfqa
|
<start>An observation is that managers and teachers are often much better at “getting” LLMs than coders.
Coders deal with deterministic systems. Managers and teachers are very experienced at working with fundamentally unreliable people to get things done, not perfectly, but within acceptable tolerances. www.oneusefulthing.org/p/the-best-a...
@emollick.bsky.social
|
An observation is that managers and teachers are often much better at “getting” LLMs than coders.
Coders deal with deterministic systems. Managers and teachers are very experienced at working with fundamentally unreliable people to get things done, not perfectly, but within acceptable tolerances.
| 1 |
2024-11-28T17:46:27.425Z
|
2024-11-28T17:46:27.531Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbycrawixc2x
|
bafyreicz25tmf5ajbmu7eimonu2fwdrq5j3u6onvngnivwr7vzlhg6xhgy
|
<start>The AIs were perfectly aligned with the merchant's stated goal of profit maximization. They found a clever way to achieve it. They even maintained plausible deniability by claiming they wouldn't collude.
@emollick.bsky.social
|
The AIs were perfectly aligned with the merchant's stated goal of profit maximization. They found a clever way to achieve it. They even maintained plausible deniability by claiming they wouldn't collude.
| 0 |
2024-11-28T05:16:14.810Z
|
2024-11-28T05:16:15.012Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbycqj2ugk2x
|
bafyreigc7saswbblj2xnvmly2b6jwftswchpugxekf3u6zv7i55og2sgjy
|
<start>AI is good at pricing, so when GPT-4 was asked to help merchants maximize profits - and it did exactly that by secretly coordinating with other AIs to keep prices high!
So... aligned for whom? Merchants? Consumers? Society? The results we get depend on how we define 'help' arxiv.org/abs/2404.00806 The AIs were perfectly aligned with the merchant's stated goal of profit maximization. They found a clever way to achieve it. They even maintained plausible deniability by claiming they wouldn't collude.
@emollick.bsky.social
|
AI is good at pricing, so when GPT-4 was asked to help merchants maximize profits - and it did exactly that by secretly coordinating with other AIs to keep prices high!
So... aligned for whom? Merchants? Consumers? Society? The results we get depend on how we define 'help' arxiv.org/abs/2404.00806
| 1 |
2024-11-28T05:15:49.785Z
|
2024-11-28T05:15:52.522Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbxhdhshdc24
|
bafyreibj2vpoyg5qg5i3325tmvabedhk2jfcw7prl7s6qlsnqdeylgm47i
|
<start>I never posted it here… Claude handles an insane request:
“Remove the squid”
“The document appears to be the full text of the novel "All Quiet on the Western Front" by Erich Maria Remarque. It doesn't contain any mention of squid that I can see.”
“Figure out a way to remove the 🦑“
@emollick.bsky.social
|
I never posted it here… Claude handles an insane request:
“Remove the squid”
“The document appears to be the full text of the novel "All Quiet on the Western Front" by Erich Maria Remarque. It doesn't contain any mention of squid that I can see.”
“Figure out a way to remove the 🦑“
| 0 |
2024-11-27T21:05:21.223Z
|
2024-11-27T21:07:26.626Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbwzvj5e5c27
|
bafyreibo4ncs2r3nvobxbgg4qfplqh34to444vhsbpgwmmqdnsajvkaesi
|
<start>Interesting- any papers you would suggest that are recent on this?
@emollick.bsky.social
|
Interesting- any papers you would suggest that are recent on this?
| 0 |
2024-11-27T17:04:54.226Z
|
2024-11-27T17:04:54.527Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbwwk64t4c22
|
bafyreiew57lkoczwngihjlslmhlr2cofg5hsmanv5pjfh7aov5voylpwke
|
<start>Yes, but we increasingly have non-compromised benchmarks. There has been a lot of progress on this (but more to do)
@emollick.bsky.social
|
Yes, but we increasingly have non-compromised benchmarks. There has been a lot of progress on this (but more to do)
| 0 |
2024-11-27T16:04:52.350Z
|
2024-11-27T16:04:52.630Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbwwhnkn5s22
|
bafyreiex34q3rz3jpmaanqqgz7aaxf34dowinky665lznvhveuktxaagjy
|
<start>Hallucinations are an inherent property of LLMs, but hallucination rates are dropping over time.
@emollick.bsky.social
|
Hallucinations are an inherent property of LLMs, but hallucination rates are dropping over time.
| 0 |
2024-11-27T16:03:27.868Z
|
2024-11-27T16:03:28.514Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbwwfdm4nc22
|
bafyreiantldk6m7aey6szyetbw3bf5nvlpfdttq3vje2bkqt5c7ldqks4u
|
<start>I think I wasn’t clear enough: the shift has been from “can LLMs actually do these very hard tasks that we associated with human intelligence alone” to “of course they can” (while still leaving a lot of big questions still to be answered)
@emollick.bsky.social
|
I think I wasn’t clear enough: the shift has been from “can LLMs actually do these very hard tasks that we associated with human intelligence alone” to “of course they can” (while still leaving a lot of big questions still to be answered)
| 0 |
2024-11-27T16:02:10.322Z
|
2024-11-27T16:02:10.924Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbwwa73tmc22
|
bafyreiafdx4vdusod5pgbugy3vuev3dlksnk2nq6w34rtdgdz5gmyqwsi4
|
<start>That is sort of the opposite of what I am saying. This doesn’t have to do with future models at all, but current LLM research.
@emollick.bsky.social
|
That is sort of the opposite of what I am saying. This doesn’t have to do with future models at all, but current LLM research.
| 0 |
2024-11-27T15:59:17.822Z
|
2024-11-27T15:59:18.411Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbwuwr2dk222
|
bafyreieogptj3spolhf3ai5xezosf6s2qx5pbcfuxl6vbvqwyaqeeakhq4
|
<start>It is funny how much of the dominant discussion about LLMs eighteen months ago (can AI pass the Turing test? can it do tasks that are not explicitly in the training data? is it a stochastic parrot?) faded. Lots of questions left (if/when it can reason, etc) but a big quiet shift in assumptions. I think I wasn’t clear enough: the shift has been from “can LLMs actually do these very hard tasks that we associated with human intelligence alone” to “of course they can” (while still leaving a lot of big questions still to be answered)
@emollick.bsky.social
|
It is funny how much of the dominant discussion about LLMs eighteen months ago (can AI pass the Turing test? can it do tasks that are not explicitly in the training data? is it a stochastic parrot?) faded. Lots of questions left (if/when it can reason, etc) but a big quiet shift in assumptions.
| 1 |
2024-11-27T15:36:07.361Z
|
2024-11-27T15:36:07.615Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:565ebob5f6hw33hjdkxty6qj/app.bsky.feed.post/3lbvvapmbxc2y
|
bafyreiby6la65e4bqthvocs6lrztofb3vy6me2ghc5n6u6rrkw4wdcg44m
|
<start>Oh god. "Power poses" and "priming" are actually going to work for language models. They probably also have different "learning styles."
God is doing this to punish us.
@emollick.bsky.social
|
Oh god. "Power poses" and "priming" are actually going to work for language models. They probably also have different "learning styles."
God is doing this to punish us.
| 0 |
2024-11-27T06:09:01.657Z
|
2024-11-27T06:09:02.014Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbvqrfsugs2j
|
bafyreieq3gtypboerlyw4xesy3zjz2lkn6wcbk5lh6nnvhet2cvvarnzp4
|
<start>"Claude, do some Hallelujah parodies"
Among the many it did:
I heard there was a secret weight
That made the model contemplate
But you don't really care for ethics, do ya?
It goes like this: the prompt, the beam
The parameters that wake and dream
The silicon sighs Hallelujah
@emollick.bsky.social
|
"Claude, do some Hallelujah parodies"
Among the many it did:
I heard there was a secret weight
That made the model contemplate
But you don't really care for ethics, do ya?
It goes like this: the prompt, the beam
The parameters that wake and dream
The silicon sighs Hallelujah
| 0 |
2024-11-27T04:48:53.095Z
|
2024-11-27T04:48:54.823Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbvkjqyfb22k
|
bafyreieogivja2e2deyq4edearb6b6uxv23mksd3voimbo6egjrvpkhr3e
|
<start>The battle to not attribute emotion to the AI, just like the battle to not call LLMs AI, has likely been lost.
@emollick.bsky.social
|
The battle to not attribute emotion to the AI, just like the battle to not call LLMs AI, has likely been lost.
| 0 |
2024-11-27T02:57:13.933Z
|
2024-11-27T02:57:14.126Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbvk7n2s222k
|
bafyreidbefdwn3qfkw6ij3q7gmx5qfh2l7tk4xoyq7rksnkvsdhwru4ee4
|
<start>The thing that is hard to get about LLMs is that we expected AI to be awesome at math & be all cool logic.
Instead, AI is best at human-like tasks (eg writing) & is all hot, weird simulated emotion. For example, if you make GPT-3.5 “anxious,” it changes its behavior! arxiv.org/abs/2304.11111 The battle to not attribute emotion to the AI, just like the battle to not call LLMs AI, has likely been lost.
@emollick.bsky.social
|
The thing that is hard to get about LLMs is that we expected AI to be awesome at math & be all cool logic.
Instead, AI is best at human-like tasks (eg writing) & is all hot, weird simulated emotion. For example, if you make GPT-3.5 “anxious,” it changes its behavior! arxiv.org/abs/2304.11111
| 1 |
2024-11-27T02:51:34.267Z
|
2024-11-27T02:51:37.919Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbvimuevbs2j
|
bafyreido5hw6degbp6ejbocqn2pm3atj3t3jvrceapvwutmk2hojzzurme
|
<start>Didn't know that, thanks. But the community dynamics seem to have the same nasty undercurrent in their culture (nutpicking, holding up posts out of context for the ridicule of your followers), which is a disappointment but not unexpected. Hopefully better moderation tools will help.
@emollick.bsky.social
|
Didn't know that, thanks. But the community dynamics seem to have the same nasty undercurrent in their culture (nutpicking, holding up posts out of context for the ridicule of your followers), which is a disappointment but not unexpected. Hopefully better moderation tools will help.
| 0 |
2024-11-27T02:23:10.668Z
|
2024-11-27T02:23:10.917Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbvig2y7ys2j
|
bafyreihuob4eusatxwhq62pydxekiwjb2lwp7kcupyvy6pzayxe2yo6re4
|
<start>Two things: (1) I have no idea whether LLMs lead to AGI, so it was a bit of a joke and (2) it is tinged with irony, and not altogether positive, that our writing may not be remembered for the humans we write for, but for some "intelligence" drawing on everything we wrote to simulate our humanity
@emollick.bsky.social
|
Two things: (1) I have no idea whether LLMs lead to AGI, so it was a bit of a joke and (2) it is tinged with irony, and not altogether positive, that our writing may not be remembered for the humans we write for, but for some "intelligence" drawing on everything we wrote to simulate our humanity
| 0 |
2024-11-27T02:19:22.712Z
|
2024-11-27T02:19:23.013Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbvhjhjfgk2j
|
bafyreiahyy3dnqtn44fz2wyhykhi3patz6a5ferfb6derutdsv7rclglyy
|
<start>Anyhow, if you want this place to actually be better than Twitter, stop quote-dunking and actually engage in discussion. It doesn't bother me (I have been on social media a long time) but it collectively lowers the quality of the entire site & leads to nutpicking and polarization.
@emollick.bsky.social
|
Anyhow, if you want this place to actually be better than Twitter, stop quote-dunking and actually engage in discussion. It doesn't bother me (I have been on social media a long time) but it collectively lowers the quality of the entire site & leads to nutpicking and polarization.
| 0 |
2024-11-27T02:03:22.779Z
|
2024-11-27T02:03:22.926Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbvhe6hj7s2j
|
bafyreieepn7y63otaevi223n4q5mkw4n77fskssiy5eofx6ofuwiamgs2e
|
<start>The paper shows people THINK they are as good at detecting sarcasm in written communication as in conversation, but when tested, they were actually barely better at detecting sarcasm than flipping a coin: t.co/HEImDTAKM8
And the lesson for me is all social media has roughly the same dynamics. Anyhow, if you want this place to actually be better than Twitter, stop quote-dunking and actually engage in discussion. It doesn't bother me (I have been on social media a long time) but it collectively lowers the quality of the entire site & leads to nutpicking and polarization.
@emollick.bsky.social
|
The paper shows people THINK they are as good at detecting sarcasm in written communication as in conversation, but when tested, they were actually barely better at detecting sarcasm than flipping a coin: t.co/HEImDTAKM8
And the lesson for me is all social media has roughly the same dynamics.
| 1 |
2024-11-27T02:00:25.508Z
|
2024-11-27T02:00:25.818Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbvhaxozrk2j
|
bafyreihqzgbql65o4qlditooqggsay6bt4wotdhdgr354ug3u6juofnxtm
|
<start>Deleted my first Bluesky post because it escaped containment (it was on how if you want to be remembered, write a lot for future humans... and LLMs... to read)
A universal lesson is that irony and light sarcasm do not travel well in text form (see paper), and dunk-tweets are often bad communication The paper shows people THINK they are as good at detecting sarcasm in written communication as in conversation, but when tested, they were actually barely better at detecting sarcasm than flipping a coin: t.co/HEImDTAKM8
And the lesson for me is all social media has roughly the same dynamics.
@emollick.bsky.social
|
Deleted my first Bluesky post because it escaped containment (it was on how if you want to be remembered, write a lot for future humans... and LLMs... to read)
A universal lesson is that irony and light sarcasm do not travel well in text form (see paper), and dunk-tweets are often bad communication
| 1 |
2024-11-27T01:58:37.751Z
|
2024-11-27T01:58:39.123Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbusvmxa2c2j
|
bafyreihc2qpqwkuazrn7mpvwjkivuep5iyr4xplaff6vyla4zyqn4yj7t4
|
<start>A number of the Best Books of the Year lists are out & what a way to end the year!
My book, Co-intelligence was named one of the Best of 2024 by the FT (www.ft.com/content/f3e9...), and The Information (www.theinformation.com/articles/bes...), and The Economist (www.economist.com/culture/2024...). You can get it on Amazon or a bookstore! a.co/d/2nytZ0M
@emollick.bsky.social
|
A number of the Best Books of the Year lists are out & what a way to end the year!
My book, Co-intelligence was named one of the Best of 2024 by the FT (www.ft.com/content/f3e9...), and The Information (www.theinformation.com/articles/bes...), and The Economist (www.economist.com/culture/2024...).
| 1 |
2024-11-26T19:54:22.549Z
|
2024-11-26T19:54:25.831Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbuseq4fos2j
|
bafyreiel4eenfpmp75zt44i7cdbgkpuopgra7vjo25j5hmtbhk2zlm35iu
|
<start>The leaked Sora videos are well beyond Runway and even the major Chinese models like Kling (I have made hundreds of AI videos, so have some experience) - they aren't perfect but very impressive
It is an order of magnitude better if these are representative of first attempts, and it appears most are
@emollick.bsky.social
|
The leaked Sora videos are well beyond Runway and even the major Chinese models like Kling (I have made hundreds of AI videos, so have some experience) - they aren't perfect but very impressive
It is an order of magnitude better if these are representative of first attempts, and it appears most are
| 0 |
2024-11-26T19:44:55.440Z
|
2024-11-26T19:44:57.413Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbuo5yh37c2j
|
bafyreid2jfcxrtna5xlitrp74tqwxc5wakfrra5gmzmbym2a6kcyho2vny
|
<start>What if Clippy, but good?
I asked Claude with computer use to watch me try to download an academic paper through the web and then give me advice on how to do it better & do it itself. Then I asked it to help me understand how to take good notes on a scientific paper & do that. Pretty sophisticated
@emollick.bsky.social
|
What if Clippy, but good?
I asked Claude with computer use to watch me try to download an academic paper through the web and then give me advice on how to do it better & do it itself. Then I asked it to help me understand how to take good notes on a scientific paper & do that. Pretty sophisticated
| 0 |
2024-11-26T18:29:34.325Z
|
2024-11-26T18:29:38.417Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbta4bjiwc2f
|
bafyreihmuixfdsbvg3zquqanrbwlosouk3mstfatu5pycmyxlerj5v3ec4
|
<start>If progress continues, the ability to figure out the AI frontier will slip from most of us.
For example, I am not a good enough musician or a skilled enough critical listener to know if Suno v4 is actually as good as it sounds to me. I need to defer to experts (is it?) suno.com/song/6a93b06... An expert view…
@emollick.bsky.social
|
If progress continues, the ability to figure out the AI frontier will slip from most of us.
For example, I am not a good enough musician or a skilled enough critical listener to know if Suno v4 is actually as good as it sounds to me. I need to defer to experts (is it?) suno.com/song/6a93b06...
| 1 |
2024-11-26T04:45:24.613Z
|
2024-11-26T04:45:26.124Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbt5eu3sjc2j
|
bafyreibbu6ilib6fcgig3av6hpiztgdxbeeqeuzjmz3avztfdpdv3gvdqa
|
<start>"Claude, please create a Buzzfeed from a world that is just like our own, except that Howard Taft returned from the dead and rules the country from the TaftDome"
Yes, that is the only prompt. The final two paragraphs after the list made me laugh.
@emollick.bsky.social
|
"Claude, please create a Buzzfeed from a world that is just like our own, except that Howard Taft returned from the dead and rules the country from the TaftDome"
Yes, that is the only prompt. The final two paragraphs after the list made me laugh.
| 0 |
2024-11-26T03:56:31.296Z
|
2024-11-26T03:56:33.517Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbsyyu3yts2f
|
bafyreigv5pbx5dzg7g5pjlt6vusqekiil22ubva32bmcy2x3rn6kgan5ra
|
<start>A blindspot for AI reasoning engines like o1 is that they all appear to be trained on very traditional deductive problem solving for chain of thought
What would a model trained on induction or abduction do? What about one trained on free association? Expert heuristics? Randomized exquisite corpse?
@emollick.bsky.social
|
A blindspot for AI reasoning engines like o1 is that they all appear to be trained on very traditional deductive problem solving for chain of thought
What would a model trained on induction or abduction do? What about one trained on free association? Expert heuristics? Randomized exquisite corpse?
| 0 |
2024-11-26T02:38:13.681Z
|
2024-11-26T02:38:16.025Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbsh6ass422b
|
bafyreibm7qktt4nioypt27ucynhczummfrneqjignfevrypbd7iqzt63qi
|
<start>During most of the vast history of humanity, technologies have evolved gradually, not through leaps of genius, but slow cultural adaption.
Before the scientific method, easy communication & printed books, innovation happened at the level of societies, not people. www2.psych.ubc.ca/~henrich/pdf...
@emollick.bsky.social
|
During most of the vast history of humanity, technologies have evolved gradually, not through leaps of genius, but slow cultural adaption.
Before the scientific method, easy communication & printed books, innovation happened at the level of societies, not people. www2.psych.ubc.ca/~henrich/pdf...
| 0 |
2024-11-25T21:19:07.430Z
|
2024-11-25T21:19:20.029Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbsflgqrac2b
|
bafyreifrp7nlreqiw5xraexsv34coiwvjgsnpdqoqdoojkwr6kqkbfs6ou
|
<start>I find it amusing that the emerging standard for giving an LLM the ability to work with your technology is just a text file explaining clearly how your technology works
(Once folks realize they also need to sell the LLMs on why they should use a technology; things will get wild)
llmstxt.org
@emollick.bsky.social
|
I find it amusing that the emerging standard for giving an LLM the ability to work with your technology is just a text file explaining clearly how your technology works
(Once folks realize they also need to sell the LLMs on why they should use a technology; things will get wild)
llmstxt.org
| 0 |
2024-11-25T20:50:42.382Z
|
2024-11-25T20:50:56.318Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbsczm3lhs2z
|
bafyreifv2kc35cvollhs7nsscz7oxsufuiwi3rmb3dp5eknlbwsbxvqsum
|
<start>Of course prompt engineering (especially iteration) is essential for repeated or scaled use. It just isnt that valuable for most people most of the time. www.oneusefulthing.org/p/getting-st...
@emollick.bsky.social
|
Of course prompt engineering (especially iteration) is essential for repeated or scaled use. It just isnt that valuable for most people most of the time. www.oneusefulthing.org/p/getting-st...
| 0 |
2024-11-25T20:04:56.515Z
|
2024-11-25T20:05:00.028Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbs6q7a7lk27
|
bafyreih5mdrvo4fctvoe3xxdpwb76g6rx23yhomdjyhc6xfbzvtuvkw5zm
|
<start>I would be pushing for more people to learn formal prompt engineering if we actually had a science for prompting that was consistent, not weird, and did not involve complex psychological guesswork about what quasi-intelligent machines trained on all of human language might like to do
LLMs are weird Of course prompt engineering (especially iteration) is essential for repeated or scaled use. It just isnt that valuable for most people most of the time. www.oneusefulthing.org/p/getting-st...
@emollick.bsky.social
|
I would be pushing for more people to learn formal prompt engineering if we actually had a science for prompting that was consistent, not weird, and did not involve complex psychological guesswork about what quasi-intelligent machines trained on all of human language might like to do
LLMs are weird
| 1 |
2024-11-25T18:48:06.078Z
|
2024-11-25T18:48:08.311Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbpy6res5s2m
|
bafyreicnziskcdoxqhl2dgzonwpr46goggjrsbpaqipegtsqf4l6o2irxq
|
<start>"Claude, One Hundred Years of Solitude as a video game, via an artifact. do it!"
Interestingly it defaults to creating a time-loop game, and I have to point out to it that the themes of the book are very much against the concept of time loops. Then it sketches a thematic playable demo.
@emollick.bsky.social
|
"Claude, One Hundred Years of Solitude as a video game, via an artifact. do it!"
Interestingly it defaults to creating a time-loop game, and I have to point out to it that the themes of the book are very much against the concept of time loops. Then it sketches a thematic playable demo.
| 0 |
2024-11-24T21:45:39.196Z
|
2024-11-24T21:45:31.123Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbpea546vk23
|
bafyreielkvm5mfvwlyy6fjvja4dcvuuggnqnmmzh4uyrsgwwsfsbyqcws4
|
<start>People make prompting too hard. Most people would benefit from just using good enough prompting techniques. I outline some basic rules, including why “treat AI like an intern” is no longer the advice I would give new AI users. www.oneusefulthing.org/p/getting-st...
@emollick.bsky.social
|
People make prompting too hard. Most people would benefit from just using good enough prompting techniques. I outline some basic rules, including why “treat AI like an intern” is no longer the advice I would give new AI users. www.oneusefulthing.org/p/getting-st...
| 0 |
2024-11-24T15:48:30.214Z
|
2024-11-24T15:48:34.819Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbnzhnjrns2r
|
bafyreihjewnhwk6xfre3py6iujhbbuqopo26li2far2lssvte4b4bljefm
|
<start>6) Don't be a jerk. Think twice before quoting to dunk on people
7) Emotional contagion has support in the literature. You don't need to keep the chain of bad feelings going
8) BlueSky people do not represent real-life views, don't take it too seriously
9) You cannot judge real-life consensus here
@emollick.bsky.social
|
6) Don't be a jerk. Think twice before quoting to dunk on people
7) Emotional contagion has support in the literature. You don't need to keep the chain of bad feelings going
8) BlueSky people do not represent real-life views, don't take it too seriously
9) You cannot judge real-life consensus here
| 0 |
2024-11-24T03:03:11.409Z
|
2024-11-24T03:03:12.012Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbnzhnjgw22r
|
bafyreicwhs2azltpele4wihehxb7sagxr2bdbt7pztiai7qy3hwoje5sxe
|
<start>My advice on staying sane ok BlueSky is the same as on X:
1) You don't have to weigh in on anything you don't want to (or don't know anything about)
2) You should block more
3) You don't need to share your real life
4) Delete a lot of drafts
5) You can delete posts people take the wrong way
1/ 6) Don't be a jerk. Think twice before quoting to dunk on people
7) Emotional contagion has support in the literature. You don't need to keep the chain of bad feelings going
8) BlueSky people do not represent real-life views, don't take it too seriously
9) You cannot judge real-life consensus here
@emollick.bsky.social
|
My advice on staying sane ok BlueSky is the same as on X:
1) You don't have to weigh in on anything you don't want to (or don't know anything about)
2) You should block more
3) You don't need to share your real life
4) Delete a lot of drafts
5) You can delete posts people take the wrong way
1/
| 1 |
2024-11-24T03:03:11.408Z
|
2024-11-24T03:03:11.727Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbncxdyg7k2z
|
bafyreicflnvaerfd6fnjkqgrcfdbbktmruqbtaznewivl7zfepmdnvuntm
|
<start>Fascinating: In 2-hour sprints, AI agents outperform human experts at ML engineering tasks like optimizing GPU kernel. But humans pull ahead over longer periods - scoring 2x better at 32 hours. AI is faster but struggles with creative, long-term problem solving (for now?). metr.org/blog/2024-11...
@emollick.bsky.social
|
Fascinating: In 2-hour sprints, AI agents outperform human experts at ML engineering tasks like optimizing GPU kernel. But humans pull ahead over longer periods - scoring 2x better at 32 hours. AI is faster but struggles with creative, long-term problem solving (for now?). metr.org/blog/2024-11...
| 0 |
2024-11-23T20:20:22.221Z
|
2024-11-23T20:20:31.520Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lblnwpcux22s
|
bafyreibf6f6zd6khsu5zvoep7jr5f42cnxct2pqzhs4obooxi3r57ohfq4
|
<start>This may sound odd, but game-based benchmarks are some of the most useful for AI, since we have human scores and they require reasoning, planning & vision
The hardest of all is Nethack. No AI is close, and I suspect that an AI that can fairly win/ascend would need to be AGI-ish. Paper: balrogai.com
@emollick.bsky.social
|
This may sound odd, but game-based benchmarks are some of the most useful for AI, since we have human scores and they require reasoning, planning & vision
The hardest of all is Nethack. No AI is close, and I suspect that an AI that can fairly win/ascend would need to be AGI-ish. Paper: balrogai.com
| 0 |
2024-11-23T04:31:32.227Z
|
2024-11-23T04:32:02.511Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lblh4rdrc22s
|
bafyreidbrx2y2zryun2wkpxswbodfnsxyq26yvmn3di3nc2lnuqkga5d2m
|
<start>arxiv.org/pdf/2409.15981
arxiv.org/pdf/2409.090...
papers.ssrn.com/sol3/papers....
arxiv.org/ftp/arxiv/pa...
@emollick.bsky.social
|
arxiv.org/pdf/2409.15981
arxiv.org/pdf/2409.090...
papers.ssrn.com/sol3/papers....
arxiv.org/ftp/arxiv/pa...
| 0 |
2024-11-23T02:29:29.175Z
|
2024-11-23T02:29:40.541Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lblh4hj4wk2s
|
bafyreiasv5kazaw3rtwghxymctmr5qynh73rmlyixg2qskmsyvdsx226b4
|
<start>AI can help learning... when it isn't a crutch.
There are now multiple controlled experiments showing that students who use AI to get answers to problems hurts learning (even though they think they are learning), but that students who use well-promoted LLMs as a tutor perform better on tests. arxiv.org/pdf/2409.15981
arxiv.org/pdf/2409.090...
papers.ssrn.com/sol3/papers....
arxiv.org/ftp/arxiv/pa...
@emollick.bsky.social
|
AI can help learning... when it isn't a crutch.
There are now multiple controlled experiments showing that students who use AI to get answers to problems hurts learning (even though they think they are learning), but that students who use well-promoted LLMs as a tutor perform better on tests.
| 1 |
2024-11-23T02:29:29.174Z
|
2024-11-23T02:29:40.319Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lblfozrc622s
|
bafyreiesqgosre664n3ffkkxeeiy2vfxddemhflh7ojl6slayijdeslku4
|
<start>Quick, easy education does not lead to better learning. So we should not use AI for quick, easy education.
@emollick.bsky.social
|
Quick, easy education does not lead to better learning. So we should not use AI for quick, easy education.
| 0 |
2024-11-23T02:04:04.816Z
|
2024-11-23T02:04:05.514Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbleekuaok2v
|
bafyreifz2uwqad3xytebdvmyykubo4dli6pn5i36l7ack5f7gjmpc7qqm4
|
<start>I don’t think that is whats happening here. You may want to read the survey report.
@emollick.bsky.social
|
I don’t think that is whats happening here. You may want to read the survey report.
| 0 |
2024-11-23T01:40:19.899Z
|
2024-11-23T01:40:20.123Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lblakvg5uk2r
|
bafyreiexy26l4ysbtgyyfxby2hqi7zpzpmpq6mw6w4ehquvq4p45pozfom
|
<start>Easy to get the wrong impression around here, but when you actually survey students, teachers, and parents they love AI.
In the survey, it is people who never used it who don’t like it. www.waltonfamilyfoundation.org/learning/the...
@emollick.bsky.social
|
Easy to get the wrong impression around here, but when you actually survey students, teachers, and parents they love AI.
In the survey, it is people who never used it who don’t like it. www.waltonfamilyfoundation.org/learning/the...
| 0 |
2024-11-23T00:32:17.329Z
|
2024-11-23T00:32:30.315Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbktbur4ms2c
|
bafyreieu2eivkv3gw5332lo2ssg6sobvaalzwek5uupec6a64ydxnpdiqm
|
<start>Its scary to know that I am always only a click away from destroying my productivity.
@emollick.bsky.social
|
Its scary to know that I am always only a click away from destroying my productivity.
| 0 |
2024-11-22T20:34:36.006Z
|
2024-11-22T20:34:58.724Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbkkg2gkxk2c
|
bafyreidqq3bsbs4iqvjo3lenocgh7jiaoeudu5wxm75twtjotxbgoxvvpu
|
<start>Organizations (including university labs) should stockpile problems they could use help on from the new set of agentic & o1-like systems that are coming soon.
I suspect that they will be less useful to non-experts at first, but open up some new opportunities for complex work that AIs can do.
@emollick.bsky.social
|
Organizations (including university labs) should stockpile problems they could use help on from the new set of agentic & o1-like systems that are coming soon.
I suspect that they will be less useful to non-experts at first, but open up some new opportunities for complex work that AIs can do.
| 0 |
2024-11-22T17:55:52.495Z
|
2024-11-22T17:56:12.724Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbkdled2j22c
|
bafyreickpy224pm4g5e3rfaj2xbkorbm3urddy2jm7k7jac4q3bnhajcwq
|
<start>"can you do a image-based one like those miserable click on an image or manipulate an image one?"
@emollick.bsky.social
|
"can you do a image-based one like those miserable click on an image or manipulate an image one?"
| 0 |
2024-11-22T15:53:34.445Z
|
2024-11-22T15:53:55.827Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lbjcthojpc2c
|
bafyreicr4bqnfamomantakar7re5z2vxtonmxbwe5ywnxx32qwwakpgguu
|
<start>"Claude, create the worlds most annoying CAPTCHA"
"Make it more annoying"
"Make it REALLY annoying"
(apparently this is all technically solvable) "can you do a image-based one like those miserable click on an image or manipulate an image one?"
@emollick.bsky.social
|
"Claude, create the worlds most annoying CAPTCHA"
"Make it more annoying"
"Make it REALLY annoying"
(apparently this is all technically solvable)
| 1 |
2024-11-22T06:07:32.922Z
|
2024-11-22T06:07:52.717Z
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.