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/3ldmh7lmit22c
|
bafyreibt6zbqgngklvefp4ikr772rx4axyhyk5mvaate6rojlx5ooyojsi
|
<start>This does use a real dataset (it doesn't collect novel data, though)
@emollick.bsky.social
|
This does use a real dataset (it doesn't collect novel data, though)
| 0 |
2024-12-18T22:54:17.141Z
|
2024-12-18T22:54:17.924Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldls7wn2ek2c
|
bafyreigladnzyolbeepwangsfos3bfvtag2orbn72l6c7ukyglgosubvce
|
<start>"Suddenly, the barbarians arrive" (that's the full prompt)
Making veo 2 videos is fascinating - the results are often delightful and it does something interesting in the first outputs. But it is not very close to a replacement for traditional video work, it is less steerable & more serendipity.
@emollick.bsky.social
|
"Suddenly, the barbarians arrive" (that's the full prompt)
Making veo 2 videos is fascinating - the results are often delightful and it does something interesting in the first outputs. But it is not very close to a replacement for traditional video work, it is less steerable & more serendipity.
| 0 |
2024-12-18T16:38:40.115Z
|
2024-12-18T16:38:42.624Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldlq2h3q4k27
|
bafyreiaodce5mwisy6upkgtf3cm6z3ydldxwnxb7nyhbu2vkaask7xxwk4
|
<start>Researchers used AI to generate 288 complete academic finance papers predicting stock returns, complete with plausible theoretical frameworks & citations. Each paper looks and reads as legit.
They did this to show how easy it now is to mass produce "credible" research. Academia isn't ready. papers.ssrn.com/sol3/papers....
@emollick.bsky.social
|
Researchers used AI to generate 288 complete academic finance papers predicting stock returns, complete with plausible theoretical frameworks & citations. Each paper looks and reads as legit.
They did this to show how easy it now is to mass produce "credible" research. Academia isn't ready.
| 1 |
2024-12-18T15:59:48.556Z
|
2024-12-18T15:59:51.830Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldkt2nikpc2c
|
bafyreihkgi3ytdsjqd2y2bf6hctdwxkuq5las45intix2a5ci62tv5gpdu
|
<start>"The camera swoops through a hyperealistic space evocative of piranesi and escher, filled with stairs and ruins, a couple holding hands walks down one of the stairs"
Impressed by the shadows, the solidity of the people's feet on the stairs, the switching of the held hands... "Suddenly, the barbarians arrive" (that's the full prompt)
Making veo 2 videos is fascinating - the results are often delightful and it does something interesting in the first outputs. But it is not very close to a replacement for traditional video work, it is less steerable & more serendipity.
@emollick.bsky.social
|
"The camera swoops through a hyperealistic space evocative of piranesi and escher, filled with stairs and ruins, a couple holding hands walks down one of the stairs"
Impressed by the shadows, the solidity of the people's feet on the stairs, the switching of the held hands...
| 1 |
2024-12-18T07:20:56.762Z
|
2024-12-18T07:20:58.122Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldkrg33z6c27
|
bafyreiflsznseaveeixgxn5sj4rjl5wmvzzwis2g4vw3ckiz5c77phaviq
|
<start>AI is not unambiguously good or bad, and it is going to have widespread effects across all of society, some for the better, some for the worst. But it isn't going away. I think that I am an AI pragmatist, thinking about how to mitigate the downside while figuring out how to use the upside to help us
@emollick.bsky.social
|
AI is not unambiguously good or bad, and it is going to have widespread effects across all of society, some for the better, some for the worst. But it isn't going away. I think that I am an AI pragmatist, thinking about how to mitigate the downside while figuring out how to use the upside to help us
| 0 |
2024-12-18T06:51:32.644Z
|
2024-12-18T06:51:33.019Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldkrbtlyak27
|
bafyreigxau3zyuwdzjlnn2rism65ckffh5typyd3meouwmwbt4dpwcuwty
|
<start>Social media is a terrible place for debate unless it is moderated
@emollick.bsky.social
|
Social media is a terrible place for debate unless it is moderated
| 0 |
2024-12-18T06:49:10.562Z
|
2024-12-18T06:49:11.119Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldkpxyhxls27
|
bafyreie2kq5tew5lmb73rtvq7wn2i4k35auumixthfe5twl6g2wjfy2xlm
|
<start>I am interested in many other things other than AI, of course, but social media is not representative of my real life, and I tend to post more about AI than other topics nowadays, because I think that is the most valuable and interesting stuff for me to share. AI is not unambiguously good or bad, and it is going to have widespread effects across all of society, some for the better, some for the worst. But it isn't going away. I think that I am an AI pragmatist, thinking about how to mitigate the downside while figuring out how to use the upside to help us
@emollick.bsky.social
|
I am interested in many other things other than AI, of course, but social media is not representative of my real life, and I tend to post more about AI than other topics nowadays, because I think that is the most valuable and interesting stuff for me to share.
| 1 |
2024-12-18T06:25:46.387Z
|
2024-12-18T06:25:46.816Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldkpuxjvxs27
|
bafyreifz6moa5wreotz5xwl3wwmllwvsccuopn2fj2g6lkemvylbw5d4v4
|
<start>Not to be difficult, but if you don't want to hear about AI stuff and research you should probably not follow me on here?
I have learned from my experience with many social sites over the years that it is better to just block people who are snide or mean in the comments rather than argue with them. I am interested in many other things other than AI, of course, but social media is not representative of my real life, and I tend to post more about AI than other topics nowadays, because I think that is the most valuable and interesting stuff for me to share.
@emollick.bsky.social
|
Not to be difficult, but if you don't want to hear about AI stuff and research you should probably not follow me on here?
I have learned from my experience with many social sites over the years that it is better to just block people who are snide or mean in the comments rather than argue with them.
| 1 |
2024-12-18T06:24:04.731Z
|
2024-12-18T06:24:05.116Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldknxzc27k25
|
bafyreiew6hpkddn36wa3kmnv35r7oeodqhbloirotrnjolitm6rjksxdjq
|
<start>I see again the out-of-context problem that haunts this site. I am obviously not saying that scams and involuntary explicit images are not a problem, I am trying to talk about the fact that I thought deepfakes would be a bigger politician problem. May have to delete this if that isn’t clear. Decided to delete. Posting here feels very fraught at times.
@emollick.bsky.social
|
I see again the out-of-context problem that haunts this site. I am obviously not saying that scams and involuntary explicit images are not a problem, I am trying to talk about the fact that I thought deepfakes would be a bigger politician problem. May have to delete this if that isn’t clear.
| 1 |
2024-12-18T05:49:59.759Z
|
2024-12-18T05:49:59.924Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldknn6vxrs25
|
bafyreibonhfgkaxp3d4dgdqjk7qqtqaewbf42uyp63mkhni2njhkiqbrne
|
<start>Folks, I purposely did not post a believable deepfake of a real person because I don’t want to spread believable deepfakes. I thought this was a fun illustration of how a real person can be added to a video with a text prompt, but it is easy to create ones of any people you like. I see again the out-of-context problem that haunts this site. I am obviously not saying that scams and involuntary explicit images are not a problem, I am trying to talk about the fact that I thought deepfakes would be a bigger politician problem. May have to delete this if that isn’t clear.
@emollick.bsky.social
|
Folks, I purposely did not post a believable deepfake of a real person because I don’t want to spread believable deepfakes. I thought this was a fun illustration of how a real person can be added to a video with a text prompt, but it is easy to create ones of any people you like.
| 1 |
2024-12-18T05:43:56.555Z
|
2024-12-18T05:43:56.719Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldknfrdqcc25
|
bafyreifimkr2fao24tkgvm2vy5lggsddapurpnggsnm22gjkfjutuinmnq
|
<start>Yes. That is right, this is not a real video because I don’t actually want to post misinformation. Real deep fakes have been easy for at least a year.
@emollick.bsky.social
|
Yes. That is right, this is not a real video because I don’t actually want to post misinformation. Real deep fakes have been easy for at least a year.
| 0 |
2024-12-18T05:39:47.445Z
|
2024-12-18T05:39:47.819Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldk75zyawc2d
|
bafyreiglo2c43y6x53muo3oa72jpaneisbr3g42xdsp3j4igwhq3n7z2om
|
<start>"a pug is crowned king of england in an elaborate ceremony, ending with a crown being placed on its head" "The camera swoops through a hyperealistic space evocative of piranesi and escher, filled with stairs and ruins, a couple holding hands walks down one of the stairs"
Impressed by the shadows, the solidity of the people's feet on the stairs, the switching of the held hands...
@emollick.bsky.social
|
"a pug is crowned king of england in an elaborate ceremony, ending with a crown being placed on its head"
| 1 |
2024-12-18T01:24:55.685Z
|
2024-12-18T01:24:46.617Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldk73zd32k2d
|
bafyreia74ybaibfjnbd2fyvpk4ra3pb4gzghgfro6qn2hqkefnly4rwzji
|
<start>Thank you! It helps that things are changing so rapidly as well!
@emollick.bsky.social
|
Thank you! It helps that things are changing so rapidly as well!
| 0 |
2024-12-18T01:23:47.883Z
|
2024-12-18T01:23:37.917Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldk2xupx322d
|
bafyreieqg6kr2g7pcbuucbneecgwagmmdgf546ixmzouolqujbpueb2ybq
|
<start>I continually get great initial results from a single prompt. It feels crazy to be able to summon these videos into existence.
"a woman stands on featureless high tower made of concrete with veins of gold and lapis, below waves crash as a stormy ocean spreads in all directions" "a pug is crowned king of england in an elaborate ceremony, ending with a crown being placed on its head"
@emollick.bsky.social
|
I continually get great initial results from a single prompt. It feels crazy to be able to summon these videos into existence.
"a woman stands on featureless high tower made of concrete with veins of gold and lapis, below waves crash as a stormy ocean spreads in all directions"
| 1 |
2024-12-18T00:09:53.877Z
|
2024-12-18T00:09:44.923Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldjl5jrpnc2l
|
bafyreigh5s2romuifejkcz6xo3a4h5afe4jfw725uaggk44h6cxucktltm
|
<start>"black and white documentary footage of the filming of the Apollo Moon Landing on a soundstage, with appropriate 1960s era equipment"
I think the lighting on the woman's face is super interesting here, as is the physicality of the interaction with the equipment. I continually get great initial results from a single prompt. It feels crazy to be able to summon these videos into existence.
"a woman stands on featureless high tower made of concrete with veins of gold and lapis, below waves crash as a stormy ocean spreads in all directions"
@emollick.bsky.social
|
"black and white documentary footage of the filming of the Apollo Moon Landing on a soundstage, with appropriate 1960s era equipment"
I think the lighting on the woman's face is super interesting here, as is the physicality of the interaction with the equipment.
| 1 |
2024-12-17T19:26:43.858Z
|
2024-12-17T19:26:34.727Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldjhkl4y3s2l
|
bafyreihjjslbl4k455ddlhk7kyccuaufrqxhimqbye52sc6wil4shxyble
|
<start>So here is a failed case. Prompting for a complex gymnastics routine at a crumbling castle, you can see that model still can't do what is the hardest AI video edge case right now - gymnastics. Tumbling + multiple occluded limbs + human anatomy pushes these models to their limits, at least for now. "black and white documentary footage of the filming of the Apollo Moon Landing on a soundstage, with appropriate 1960s era equipment"
I think the lighting on the woman's face is super interesting here, as is the physicality of the interaction with the equipment.
@emollick.bsky.social
|
So here is a failed case. Prompting for a complex gymnastics routine at a crumbling castle, you can see that model still can't do what is the hardest AI video edge case right now - gymnastics. Tumbling + multiple occluded limbs + human anatomy pushes these models to their limits, at least for now.
| 1 |
2024-12-17T18:22:26.515Z
|
2024-12-17T18:22:17.426Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldjgj2mo3s2l
|
bafyreifkohqptmmqmekb74eulup5nwsfgpromzpva5sai5nkoit235uyba
|
<start>Yeah, prompting for particular shot types is hit-or-miss,
I think because of the short video length
@emollick.bsky.social
|
Yeah, prompting for particular shot types is hit-or-miss,
I think because of the short video length
| 0 |
2024-12-17T18:03:41.907Z
|
2024-12-17T18:03:31.728Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldjggtin3k2l
|
bafyreidxibtq6s3rkclheetqpe52imtjp6glibinvohnlbdtfdbkn5gf2y
|
<start>o1-pro is best, if you have it, for almost anything that Claude is not best for
@emollick.bsky.social
|
o1-pro is best, if you have it, for almost anything that Claude is not best for
| 0 |
2024-12-17T18:02:27.327Z
|
2024-12-17T18:02:17.215Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldjfvj4oes2l
|
bafyreifzwwf7zwyenq64vyqeondjd6y6o4wz5obermxni5zs5p57fd5rva
|
<start>‼️"o1-preview demonstrates superhuman performance in differential diagnosis, diagnostic clinical reasoning, and management reasoning, superior in multiple domains compared to prior model generations and human physicians."
And this is using vignettes, not multiple choice. arxiv.org/pdf/2412.10849
@emollick.bsky.social
|
‼️"o1-preview demonstrates superhuman performance in differential diagnosis, diagnostic clinical reasoning, and management reasoning, superior in multiple domains compared to prior model generations and human physicians."
And this is using vignettes, not multiple choice. arxiv.org/pdf/2412.10849
| 0 |
2024-12-17T17:52:46.017Z
|
2024-12-17T17:52:41.717Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldjdevlk222l
|
bafyreic62rafykjp3ibjgyg5rn2epsoopnoh7o4isljcetlmqmtf6ojo6q
|
<start>"A commander of a space ship, wearing a futuristic military uniform with a mechnical parrot on her shoulder; walks around giving orders on a damaged and burning starship bridge" So here is a failed case. Prompting for a complex gymnastics routine at a crumbling castle, you can see that model still can't do what is the hardest AI video edge case right now - gymnastics. Tumbling + multiple occluded limbs + human anatomy pushes these models to their limits, at least for now.
@emollick.bsky.social
|
"A commander of a space ship, wearing a futuristic military uniform with a mechnical parrot on her shoulder; walks around giving orders on a damaged and burning starship bridge"
| 1 |
2024-12-17T17:07:41.185Z
|
2024-12-17T17:07:32.519Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldiayqmxzc27
|
bafyreiaxfqt24plc7bhxvaqrece3oyekuney7xkskkryuu2slg7tihoefa
|
<start>"a low angle pov drone shot flies along a beach as a vast beast ascends from the water" "A commander of a space ship, wearing a futuristic military uniform with a mechnical parrot on her shoulder; walks around giving orders on a damaged and burning starship bridge"
@emollick.bsky.social
|
"a low angle pov drone shot flies along a beach as a vast beast ascends from the water"
| 1 |
2024-12-17T06:52:26.113Z
|
2024-12-17T06:52:29.917Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldiaecsx522c
|
bafyreifia6hpefizclmfzu5cs77umbyh2jaoei3fzgorejcixrj45lcew4
|
<start>"a man holding the leash of a golden retriever stares mournfully at a fireworks display over his small town of Tudor-style homes, the flashes punctuate the darkness."
All videos are from the first 4 from the prompt. I did learn that you can't ask for many cuts or scene changes. "a low angle pov drone shot flies along a beach as a vast beast ascends from the water"
@emollick.bsky.social
|
"a man holding the leash of a golden retriever stares mournfully at a fireworks display over his small town of Tudor-style homes, the flashes punctuate the darkness."
All videos are from the first 4 from the prompt. I did learn that you can't ask for many cuts or scene changes.
| 1 |
2024-12-17T06:41:00.540Z
|
2024-12-17T06:41:04.021Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldi7t4kfis2c
|
bafyreidj4zrlcv6eammon3yzsudf2vylj27lps5qczjo2e6q4wjdl2aslu
|
<start>"an woman with short black hair assembles an impossibly complicated device, close up on her face, she is sweating"
The consistency of small details is really impressive, the fact that the shaft of the screw turns at the same speed and direction, hair and sweat, tattoos... "a man holding the leash of a golden retriever stares mournfully at a fireworks display over his small town of Tudor-style homes, the flashes punctuate the darkness."
All videos are from the first 4 from the prompt. I did learn that you can't ask for many cuts or scene changes.
@emollick.bsky.social
|
"an woman with short black hair assembles an impossibly complicated device, close up on her face, she is sweating"
The consistency of small details is really impressive, the fact that the shaft of the screw turns at the same speed and direction, hair and sweat, tattoos...
| 1 |
2024-12-17T06:31:23.544Z
|
2024-12-17T06:31:27.724Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldi5rkd7hs27
|
bafyreibj55c3k2rmpp2pr763dlwf7pbuns7z6flabf7yiv2ttu26etsese
|
<start>Veo 2 prompt: "a distant shot zooms in to reveal a knight wearing a golden helmet, he begins to charge on his zebra, lowering his lance, charging towards a clockwork octopus" (this is one of the initial 4 videos it made) "an woman with short black hair assembles an impossibly complicated device, close up on her face, she is sweating"
The consistency of small details is really impressive, the fact that the shaft of the screw turns at the same speed and direction, hair and sweat, tattoos...
@emollick.bsky.social
|
Veo 2 prompt: "a distant shot zooms in to reveal a knight wearing a golden helmet, he begins to charge on his zebra, lowering his lance, charging towards a clockwork octopus" (this is one of the initial 4 videos it made)
| 1 |
2024-12-17T05:54:43.395Z
|
2024-12-17T05:54:47.519Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldi2ggehzk2c
|
bafyreib6r4chaktbdw23pc6sdhwwahnzolknewnanbkdd7nrbzbqkse2hq
|
<start>We have seen the steam turbines, the electric range finders, the concept of director firing, the plans of the Satsuma, but no one has built the fast all-big-gun battleship yet.
"claude build this idea out for me for a bit"
Yep.
@emollick.bsky.social
|
We have seen the steam turbines, the electric range finders, the concept of director firing, the plans of the Satsuma, but no one has built the fast all-big-gun battleship yet.
"claude build this idea out for me for a bit"
Yep.
| 0 |
2024-12-17T04:54:48.918Z
|
2024-12-17T04:54:53.323Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldi2ew7cnc2c
|
bafyreiggotvl6xkxqlu6gzqbjapdzeb2yblbardosseguwspnfuhrh5nui
|
<start>Feels like the Dreadnought Moment for AI is coming
When the HMS Dreadnought launched, she rendered all other warships obsolete. It was not one thing, but a mix of technologies combined into the right design that disrupted.
The pieces for transformative AI applications are there, not assembled yet. We have seen the steam turbines, the electric range finders, the concept of director firing, the plans of the Satsuma, but no one has built the fast all-big-gun battleship yet.
"claude build this idea out for me for a bit"
Yep.
@emollick.bsky.social
|
Feels like the Dreadnought Moment for AI is coming
When the HMS Dreadnought launched, she rendered all other warships obsolete. It was not one thing, but a mix of technologies combined into the right design that disrupted.
The pieces for transformative AI applications are there, not assembled yet.
| 1 |
2024-12-17T04:53:58.410Z
|
2024-12-17T04:54:04.023Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldi2bzxin222
|
bafyreiciyymxqdz6q2nhyemdqewydmew7bzqme6jiuhjb3qsa2gxmz3vxi
|
<start>Okay, veo2 from Google is pretty impressive. "Otter on a plane using wifi" produced both of these in the first batch of four (the others weren't bad either). Love the way they tap the screen.
Sora offers a lot more control options & longer clips, so it is hard to compare, but they are the big 2 now
@emollick.bsky.social
|
Okay, veo2 from Google is pretty impressive. "Otter on a plane using wifi" produced both of these in the first batch of four (the others weren't bad either). Love the way they tap the screen.
Sora offers a lot more control options & longer clips, so it is hard to compare, but they are the big 2 now
| 0 |
2024-12-17T04:52:21.691Z
|
2024-12-17T04:52:25.518Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldheydvjys27
|
bafyreibe7f6mdzpnzc5b3qn7aoipu3gslbmpfybvvlxbk6vs5jdgeewfim
|
<start>Preprint: osf.io/preprints/ps...
Paper: www.tandfonline.com/doi/full/10....
@emollick.bsky.social
|
Preprint: osf.io/preprints/ps...
Paper: www.tandfonline.com/doi/full/10....
| 0 |
2024-12-16T22:31:07.991Z
|
2024-12-16T22:31:10.127Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldherkzosk27
|
bafyreibiaj2jbrr5xd47b4ibdodvradxkrjge7233skfenogzzhizybct4
|
<start>More creative humans generate more creative output from AI image generators, as judged by other people.
The effects are highly significant, but relatively modest, suggesting that human creativity matters in getting creative work from AI... but perhaps not as might be expected. Preprint: osf.io/preprints/ps...
Paper: www.tandfonline.com/doi/full/10....
@emollick.bsky.social
|
More creative humans generate more creative output from AI image generators, as judged by other people.
The effects are highly significant, but relatively modest, suggesting that human creativity matters in getting creative work from AI... but perhaps not as might be expected.
| 1 |
2024-12-16T22:27:20.576Z
|
2024-12-16T22:27:26.226Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldh336sm3c2c
|
bafyreibzxkm3oqka2m2g3tvjswapmyvxmx37b4lu23jypsvpfyqflleqbu
|
<start>Imagen 3 is really solid. These are all from the first set of images generated
@emollick.bsky.social
|
Imagen 3 is really solid. These are all from the first set of images generated
| 0 |
2024-12-16T19:33:45.865Z
|
2024-12-16T19:33:50.324Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldgw4alk222c
|
bafyreidln6xnurfgsdblm3h5tgioropj6lhtpx7lmw64iuww3vcuoan65y
|
<start>Extra credit for the ability to differentiate otter species (again, this is one of the first four, like the other images)
@emollick.bsky.social
|
Extra credit for the ability to differentiate otter species (again, this is one of the first four, like the other images)
| 0 |
2024-12-16T18:04:52.599Z
|
2024-12-16T18:04:55.830Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldgvuj6s422c
|
bafyreibfdanvziy6d2viibgioh3a5keqwcjtreffigqrhmluo4dc6qiscq
|
<start>I have found Sora to be a very good video generation tool, but all the AI video generation tools produce very weird outputs some of the time.
This "otter on a plane using wifi" is particularly unnerving, especially as it starts so well.
@emollick.bsky.social
|
I have found Sora to be a very good video generation tool, but all the AI video generation tools produce very weird outputs some of the time.
This "otter on a plane using wifi" is particularly unnerving, especially as it starts so well.
| 0 |
2024-12-16T18:00:33.192Z
|
2024-12-16T18:00:36.917Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldgv5va53s2c
|
bafyreicok5pbhxmxan7vgxurxkrj2woesk5n4zp7iprqhixko7xigdtzmi
|
<start>With the new Imagen 3 model, I think Google's ImageFX is now a first choice image generator for most people doing causal image generation (at least this week) - really high quality images, and the interface makes prompting & playing pretty easy.
I usually get something good in 1 of the first 4 🦦 Extra credit for the ability to differentiate otter species (again, this is one of the first four, like the other images)
@emollick.bsky.social
|
With the new Imagen 3 model, I think Google's ImageFX is now a first choice image generator for most people doing causal image generation (at least this week) - really high quality images, and the interface makes prompting & playing pretty easy.
I usually get something good in 1 of the first 4 🦦
| 1 |
2024-12-16T17:47:54.058Z
|
2024-12-16T17:47:58.629Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldgu5goy5s2c
|
bafyreigvo6bw2x2hkzrqycjw7ayj7txcex7klhzbtnahdzosnez4nthspi
|
<start>Reminder that these are written as code, not as images, by models that can't "see" the results.
@emollick.bsky.social
|
Reminder that these are written as code, not as images, by models that can't "see" the results.
| 0 |
2024-12-16T17:29:45.084Z
|
2024-12-16T17:29:46.823Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldfnmqaj6222
|
bafyreicdqb2v7nfeslyyafqdudewjk6y2qpzrscwrswaxd2fewwv7afx3i
|
<start>On social media, I tend to test AI's further limits, but it does raise the usual question: if AI can't do something, is it a prompt issue or an AI issue or something else? I think most people tend to very much underestimate what AI can do.
Also, everything I post should be repeatable by anyone.
@emollick.bsky.social
|
On social media, I tend to test AI's further limits, but it does raise the usual question: if AI can't do something, is it a prompt issue or an AI issue or something else? I think most people tend to very much underestimate what AI can do.
Also, everything I post should be repeatable by anyone.
| 0 |
2024-12-16T06:00:22.481Z
|
2024-12-16T06:00:23.419Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldfnevalok22
|
bafyreic3mfc6geqbl67mh6d4j4hx6bw4nc4rcoghws4lojyzcf2titcecm
|
<start>On social media (as opposed to papers) I try to show the typical frontier of what AI can and can't do.
I don't post every failure, but I also don't generally post successes that required prompt engineering or repeated attempts, and try to be clear when I do. So selection but not cherry-picking? On social media, I tend to test AI's further limits, but it does raise the usual question: if AI can't do something, is it a prompt issue or an AI issue or something else? I think most people tend to very much underestimate what AI can do.
Also, everything I post should be repeatable by anyone.
@emollick.bsky.social
|
On social media (as opposed to papers) I try to show the typical frontier of what AI can and can't do.
I don't post every failure, but I also don't generally post successes that required prompt engineering or repeated attempts, and try to be clear when I do. So selection but not cherry-picking?
| 1 |
2024-12-16T05:55:59.291Z
|
2024-12-16T05:55:59.920Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldfk7cnhes2c
|
bafyreiap7p3ivtqb7ql4wd2ogl6ynaboetempaqsat2vocvyqhasamtgee
|
<start> All were first attempts, except for o1-pro which produced an initial error, which it fixed immediately. Reminder that these are written as code, not as images, by models that can't "see" the results.
@emollick.bsky.social
|
All were first attempts, except for o1-pro which produced an initial error, which it fixed immediately.
| 1 |
2024-12-16T04:59:10.791Z
|
2024-12-16T04:59:11.622Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldfjydv5vs2c
|
bafyreicqk7ortw4zlip4ok5je4cf73fkigq465ayld6nhkj2m3csnt527a
|
<start>I gave most of the frontier models this prompt: "create something I can paste into p5js that will startle me with its cleverness in creating something that invokes the control panel of a starship in the distant future"
Here is what they did (I didn't include Gemini 1.5, it kept making errors) All were first attempts, except for o1-pro which produced an initial error, which it fixed immediately.
@emollick.bsky.social
|
I gave most of the frontier models this prompt: "create something I can paste into p5js that will startle me with its cleverness in creating something that invokes the control panel of a starship in the distant future"
Here is what they did (I didn't include Gemini 1.5, it kept making errors)
| 1 |
2024-12-16T04:55:17.211Z
|
2024-12-16T04:55:18.720Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldfejfxtck2c
|
bafyreiahf5h2n5b5a6ykc4wj5tedfll7stnkgdvn3r3bdqbcwslcubvj2u
|
<start>Also no one is exactly sure of how they do what they do, the extent to which they reason outside of training data, how good LLMs can get, whether scaling is enough to get there, or which other approaches to take to address the issues LLMs have.
@emollick.bsky.social
|
Also no one is exactly sure of how they do what they do, the extent to which they reason outside of training data, how good LLMs can get, whether scaling is enough to get there, or which other approaches to take to address the issues LLMs have.
| 0 |
2024-12-16T03:17:27.370Z
|
2024-12-16T03:17:27.726Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldfeeaju3c2c
|
bafyreieyktcop3rx7lskfhzaitqdhpt57rmppxv6jhnrr7tsktbnjvsf6q
|
<start>Just a reminder that none of the people who make LLMs, no matter how smart, actually know what specific tasks LLMs will be good or bad at. We are barely benchmarking these systems at all on any sorts of tasks.
You should explore in areas of your expertise to try to figure it out for your use cases. Also no one is exactly sure of how they do what they do, the extent to which they reason outside of training data, how good LLMs can get, whether scaling is enough to get there, or which other approaches to take to address the issues LLMs have.
@emollick.bsky.social
|
Just a reminder that none of the people who make LLMs, no matter how smart, actually know what specific tasks LLMs will be good or bad at. We are barely benchmarking these systems at all on any sorts of tasks.
You should explore in areas of your expertise to try to figure it out for your use cases.
| 1 |
2024-12-16T03:14:33.898Z
|
2024-12-16T03:14:34.217Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldeqzfti7k2c
|
bafyreifphy2q52tcdr47fqcw64ccfl6dcq7gfylvza3xreizmbra76yxde
|
<start>We benchmark humans all the time. You measure distributions across multiple attempts, just as you do with people.
@emollick.bsky.social
|
We benchmark humans all the time. You measure distributions across multiple attempts, just as you do with people.
| 0 |
2024-12-15T21:28:29.263Z
|
2024-12-15T21:28:31.323Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldej77idac2q
|
bafyreigt3nkyagd2cxwedir5wxvrmxzzjnkccfcngqiqhu3gr7ciz3iqui
|
<start>Paper if you want to try to experiment yourself (not yet updated to correct the error): www.sciencedirect.com/science/arti...
@emollick.bsky.social
|
Paper if you want to try to experiment yourself (not yet updated to correct the error): www.sciencedirect.com/science/arti...
| 0 |
2024-12-15T19:08:33.998Z
|
2024-12-15T19:08:35.321Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldea3wgcok2u
|
bafyreibldnjzao6xogdmgamggqmzbfehbaozvxmi4jm3plit76xjpmgplu
|
<start>This was o1, not pro. I just pasted in the article with the literal prompt above.
Claude did not spot the error when given the PDF until it was told to look just at the reference value. Paper if you want to try to experiment yourself (not yet updated to correct the error): www.sciencedirect.com/science/arti...
@emollick.bsky.social
|
This was o1, not pro. I just pasted in the article with the literal prompt above.
Claude did not spot the error when given the PDF until it was told to look just at the reference value.
| 1 |
2024-12-15T16:25:40.155Z
|
2024-12-15T16:25:40.415Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lde7vt77hc2u
|
bafyreialbkkjpdkvjngyxpsjcfuo7ildw46agxs7k6qdnfmtrmqmeyjotq
|
<start>Article on error with link to paper: nationalpost.com/news/canada/... This was o1, not pro. I just pasted in the article with the literal prompt above.
Claude did not spot the error when given the PDF until it was told to look just at the reference value.
@emollick.bsky.social
|
Article on error with link to paper: nationalpost.com/news/canada/...
| 1 |
2024-12-15T16:22:15.449Z
|
2024-12-15T16:22:17.124Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lde7uhyyyc2u
|
bafyreihugqwnfifxbxmxvruq22qzdue7fk73rgu2md5zzeluztjjtxrs2e
|
<start>👀 A 10 page paper caused a recent panic because of a math error. I was curious if AI would spot the error by just prompting: “carefully check the math in this paper” especially as the info is not in training data.
o1 gets it in one shot. This feels like a big capability gain for scientific work Article on error with link to paper: nationalpost.com/news/canada/...
@emollick.bsky.social
|
👀 A 10 page paper caused a recent panic because of a math error. I was curious if AI would spot the error by just prompting: “carefully check the math in this paper” especially as the info is not in training data.
o1 gets it in one shot. This feels like a big capability gain for scientific work
| 1 |
2024-12-15T16:21:30.157Z
|
2024-12-15T16:21:32.325Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldd6yvdjus2c
|
bafyreihsi264d7ksxvukahx44nrvvtuaf6ttnwa42vf5kh5gmsozvfucie
|
<start>The calculation I have seen is 0.004kWh per request for an old-school GPT-3.5 class model, for what it is worth. arxiv.org/pdf/2304.03271
@emollick.bsky.social
|
The calculation I have seen is 0.004kWh per request for an old-school GPT-3.5 class model, for what it is worth. arxiv.org/pdf/2304.03271
| 0 |
2024-12-15T06:33:24.872Z
|
2024-12-15T06:33:25.719Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldd67pynvs2c
|
bafyreig4heo45ntiqf2eih4zqb6xsizreahgehlqz57ef7uaffkifnb7ha
|
<start>Cloud usage is more efficient than my PC. We both agree, however, that energy costs are a concern in aggregate, but a lot depends on how useful AI will be. Datacenters overall make up 1% of US energy use (this does not just include AI) - the question is how fast it will grow & how it will be powered
@emollick.bsky.social
|
Cloud usage is more efficient than my PC. We both agree, however, that energy costs are a concern in aggregate, but a lot depends on how useful AI will be. Datacenters overall make up 1% of US energy use (this does not just include AI) - the question is how fast it will grow & how it will be powered
| 0 |
2024-12-15T06:19:20.411Z
|
2024-12-15T06:19:21.222Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldd2mxatbs2c
|
bafyreifr4icbylk7kjjhmos4mo6njxn6iswzsjssiusj2yh7qj7xbe6n54
|
<start>Training takes a lot of power, but actual use is often misunderstood. If you used a frontier model until you hit your daily usage limits, it would be equivalent to a couple hours of a lightbulb being turned on.
@emollick.bsky.social
|
Training takes a lot of power, but actual use is often misunderstood. If you used a frontier model until you hit your daily usage limits, it would be equivalent to a couple hours of a lightbulb being turned on.
| 0 |
2024-12-15T05:15:09.259Z
|
2024-12-15T05:15:09.925Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldczb45ot22c
|
bafyreier5fpabwa77q5vi3gptunxccz6alu2nqrsggdzlyhbpil3wj5gku
|
<start>There have been a few papers benchmarking that. Although they are different problems depending on whether you give it the paper or expect to retrieve it from "memory"
@emollick.bsky.social
|
There have been a few papers benchmarking that. Although they are different problems depending on whether you give it the paper or expect to retrieve it from "memory"
| 0 |
2024-12-15T04:50:38.005Z
|
2024-12-15T04:50:38.520Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldcz5rmmy22c
|
bafyreieozuzwntvz2mm5rl42cm3ebolgxwjvjekhjodnberydbwgfwirga
|
<start>Very cool that I can get a GPT-4 class model running locally on my gaming computer. It passes the "rhyming poem involving cheese puns" benchmark with only a couple of strained puns.
(Seriously, it took less than 2 years to go from ChatGPT-3.5 to having a more advanced open model running at home)
@emollick.bsky.social
|
Very cool that I can get a GPT-4 class model running locally on my gaming computer. It passes the "rhyming poem involving cheese puns" benchmark with only a couple of strained puns.
(Seriously, it took less than 2 years to go from ChatGPT-3.5 to having a more advanced open model running at home)
| 0 |
2024-12-15T04:48:46.295Z
|
2024-12-15T04:48:47.923Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldcwoyfngs22
|
bafyreiblfrkede5wprav4rmunrbin435lbi2nex2bdzb22m6e6my3afp3i
|
<start>Google’s LearnLM paper for
Education and Anthropic’s paper on benchmarking overall are good places
@emollick.bsky.social
|
Google’s LearnLM paper for
Education and Anthropic’s paper on benchmarking overall are good places
| 0 |
2024-12-15T04:04:42.608Z
|
2024-12-15T04:04:42.724Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldcwngpsgs22
|
bafyreicv65fszmmp6rh3scfnb7eyho7dfu4pjph2p7tfjajo6muvvjjs3a
|
<start>And the critical missing element on the benchmarks we do have is what the average and expert human performance is at that benchmark. You need to know when to go to AI over the Best Available Human.
@emollick.bsky.social
|
And the critical missing element on the benchmarks we do have is what the average and expert human performance is at that benchmark. You need to know when to go to AI over the Best Available Human.
| 0 |
2024-12-15T04:03:50.511Z
|
2024-12-15T04:03:50.722Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldcvzvu35s22
|
bafyreigs33j2d5sjxwqfeghma2gepxuhlskevt3eu2cadfp3onbkxozf54
|
<start>I really wish there was more interest in benchmarking AI models on real tasks.
The fact that there are not 30 different benchmarks from different organizations in medicine, in law, in advice quality, etc. is a big shame. People are using systems for these things anyway & we don’t know implications. And the critical missing element on the benchmarks we do have is what the average and expert human performance is at that benchmark. You need to know when to go to AI over the Best Available Human.
@emollick.bsky.social
|
I really wish there was more interest in benchmarking AI models on real tasks.
The fact that there are not 30 different benchmarks from different organizations in medicine, in law, in advice quality, etc. is a big shame. People are using systems for these things anyway & we don’t know implications.
| 1 |
2024-12-15T03:52:55.292Z
|
2024-12-15T03:52:55.517Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldcehyoz422c
|
bafyreih7qhu374ursvhttmfn35fyonkorbzqodow56wi4moviz625pzohq
|
<start>I would say o1-pro expands the "intellectual" capabilities of AI by a lot, in quite narrow areas.
Gemini (at least in Flash form) and Llama expand where smart models and be used, and dramatically lower cost, without necessarily increasing the capability frontier beyond the current one.
@emollick.bsky.social
|
I would say o1-pro expands the "intellectual" capabilities of AI by a lot, in quite narrow areas.
Gemini (at least in Flash form) and Llama expand where smart models and be used, and dramatically lower cost, without necessarily increasing the capability frontier beyond the current one.
| 0 |
2024-12-14T22:38:40.681Z
|
2024-12-14T22:38:43.816Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldcef4oqck2c
|
bafyreidvusxc7jbhfhxl3npz4vfgqsxeveawekcwkrs3dhmp7inoip3tpi
|
<start>So these past 8 days in AI saw:
-The public release of the first post-ronnaFLOPS scale model (probably) in Gemini 2.0 Flash
-The public release of the first unconstrained test-time compute model in o1-pro
-The first small (and open!) model capable of GPT-4 level performance in Llama 3.3 I would say o1-pro expands the "intellectual" capabilities of AI by a lot, in quite narrow areas.
Gemini (at least in Flash form) and Llama expand where smart models and be used, and dramatically lower cost, without necessarily increasing the capability frontier beyond the current one.
@emollick.bsky.social
|
So these past 8 days in AI saw:
-The public release of the first post-ronnaFLOPS scale model (probably) in Gemini 2.0 Flash
-The public release of the first unconstrained test-time compute model in o1-pro
-The first small (and open!) model capable of GPT-4 level performance in Llama 3.3
| 1 |
2024-12-14T22:37:04.203Z
|
2024-12-14T22:37:06.933Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldadiaghmk2c
|
bafyreifhhr42nofyhqb4bfnkhk3dceke52azuwz4januz4qrzyvijbrzpm
|
<start>So one writes research reports and the other summarizes your documents and turns them into podcasts? Got it.
@emollick.bsky.social
|
So one writes research reports and the other summarizes your documents and turns them into podcasts? Got it.
| 0 |
2024-12-14T03:15:35.571Z
|
2024-12-14T03:15:36.722Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ldadbn3lj22c
|
bafyreidmsdjrstu46kn4dtlygfx3z4ngfbc6f7dlcfor4jtawme65tl3d4
|
<start>Google has a knack for making non-chatbot interfaces for serious work with LLMs.
When I demo them, both NotebookLM & Deep Research are instantly understandable and fill real organizational needs. They represent a tiny range of AI capability, but they are easy for everyone to get and use right away. So one writes research reports and the other summarizes your documents and turns them into podcasts? Got it.
@emollick.bsky.social
|
Google has a knack for making non-chatbot interfaces for serious work with LLMs.
When I demo them, both NotebookLM & Deep Research are instantly understandable and fill real organizational needs. They represent a tiny range of AI capability, but they are easy for everyone to get and use right away.
| 1 |
2024-12-14T03:11:53.965Z
|
2024-12-14T03:11:55.416Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ld7hlqwiuc2a
|
bafyreibyowecnpdibh7ogyiq3w6roewqrqz2d4tsnygsqjasp6g43pjyxa
|
<start>A whole bunch of systems that depend on effort being costly are going to be breaking.
Academic journals are seeing this happen already.
@emollick.bsky.social
|
A whole bunch of systems that depend on effort being costly are going to be breaking.
Academic journals are seeing this happen already.
| 0 |
2024-12-13T18:56:28.765Z
|
2024-12-13T18:56:31.321Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ld5ypvsovs2h
|
bafyreihzu74qnw7rccpndz7znex45rpifdumiv6kh5poctap3wofakvz34
|
<start>The initial invention of chess computers boosted the ability of expert human players as they learned to play better by learning from machines
But, contrary to expectations, the better chess AIs of the deep learning era, like Stockfish, have not had the same positive impact. arxiv.org/pdf/2412.03963
@emollick.bsky.social
|
The initial invention of chess computers boosted the ability of expert human players as they learned to play better by learning from machines
But, contrary to expectations, the better chess AIs of the deep learning era, like Stockfish, have not had the same positive impact. arxiv.org/pdf/2412.03963
| 0 |
2024-12-13T04:57:42.231Z
|
2024-12-13T04:57:44.423Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ld5wxonmzc2h
|
bafyreib3hvxftpegjodcx66ntjv24ysxg437ywirbrie2lwzws6sb4ynti
|
<start>Don't worry everyone, I have figured out a way to delay any potential AI takeover indefinitely. If you know you know. (Turn sound on)
@emollick.bsky.social
|
Don't worry everyone, I have figured out a way to delay any potential AI takeover indefinitely. If you know you know. (Turn sound on)
| 0 |
2024-12-13T04:26:15.682Z
|
2024-12-13T04:26:16.522Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ld53inf7rs2t
|
bafyreiej55llnwwgogxdotmbroxjy4clqqdizn6kfng3bfmvme3ozzua6q
|
<start>Did you try o1 for a research question? Curious about what you found.
@emollick.bsky.social
|
Did you try o1 for a research question? Curious about what you found.
| 0 |
2024-12-12T20:14:40.012Z
|
2024-12-12T20:14:40.329Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ld4y7zh2ss2x
|
bafyreif24jimyrutyx26akcs6kaabic774ru3nhdi4weqetscmdal5ci4q
|
<start>That isn't true, though. Novel recombination is the main pathway to innovation. And that leaves aside the ongoing debate over extrapolation.
@emollick.bsky.social
|
That isn't true, though. Novel recombination is the main pathway to innovation. And that leaves aside the ongoing debate over extrapolation.
| 0 |
2024-12-12T19:16:09.440Z
|
2024-12-12T19:16:10.534Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ld4vf7sktk2c
|
bafyreih7yn4ytrhujxoisht5qusnzepbf64bfchcquin2fl3763xpywvdm
|
<start>You don’t need to trust me. Give it a PhD problem in your field and see what happens.
It is especially good at methodologies, hypothesizing, building proofs & anything else that involves logical step-by-step thinking. It is not Google, so it won’t have recall of primary sources.
@emollick.bsky.social
|
You don’t need to trust me. Give it a PhD problem in your field and see what happens.
It is especially good at methodologies, hypothesizing, building proofs & anything else that involves logical step-by-step thinking. It is not Google, so it won’t have recall of primary sources.
| 0 |
2024-12-12T18:25:22.655Z
|
2024-12-12T18:25:23.322Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ld4tuh63522c
|
bafyreibbffiweoak7no7jctetmty36jvhaqfqtr6k6akzuz4lxiglmzp3u
|
<start>Been using o1-pro but the $20 o1 gets you most of the way there. You don’t need to trust me. Give it a PhD problem in your field and see what happens.
It is especially good at methodologies, hypothesizing, building proofs & anything else that involves logical step-by-step thinking. It is not Google, so it won’t have recall of primary sources.
@emollick.bsky.social
|
Been using o1-pro but the $20 o1 gets you most of the way there.
| 1 |
2024-12-12T17:58:06.205Z
|
2024-12-12T17:58:06.824Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ld4to3m4ts2c
|
bafyreic32ngr3nlmgut27rvevki43xckpjg56z6ac32kx2nmbnigs7t45e
|
<start>I have now run enough o1 experiments with fellow academics on their hard problems in fields ranging from neurology to genetics to economics that I believe that it has genuine potential to help with science & academics should try it to see how it does for them.
Even when it is off, it spurs thinking Been using o1-pro but the $20 o1 gets you most of the way there.
@emollick.bsky.social
|
I have now run enough o1 experiments with fellow academics on their hard problems in fields ranging from neurology to genetics to economics that I believe that it has genuine potential to help with science & academics should try it to see how it does for them.
Even when it is off, it spurs thinking
| 1 |
2024-12-12T17:54:32.755Z
|
2024-12-12T17:54:33.618Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ld3gt6lapk2e
|
bafyreia4twtjjmqqsuouo572yamxwldz7egb5md3chgywojhtbe3gggoo4
|
<start>Everyone loves this game. GOTY. Featuring 4 different minimaps!
@emollick.bsky.social
|
Everyone loves this game.
| 1 |
2024-12-12T04:32:05.262Z
|
2024-12-12T04:32:04.121Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ld3fqsktms2e
|
bafyreiag2qhw43g57igmdhnrccbvb7mdnlba2hqeijulhxo5d3e5y2vb5q
|
<start>Sora videos ripped from our collective unconscious: "Youtube video game review featuring a potato-themed FPS video game."
Sorry about this.
But good character consistency for the host across scenes, including his terrible tattoo Everyone loves this game.
@emollick.bsky.social
|
Sora videos ripped from our collective unconscious: "Youtube video game review featuring a potato-themed FPS video game."
Sorry about this.
But good character consistency for the host across scenes, including his terrible tattoo
| 1 |
2024-12-12T04:12:51.816Z
|
2024-12-12T04:12:49.816Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ld2uo42sic26
|
bafyreigyysqxhds7cbdpvj5ardkqm6fn2i2mjxf4nxfe2luzsrx3gh4khq
|
<start>To be clear, not all the effects will be good, some will he bad and some will just be very weird. We haven’t even started to grapple with long-term impacts
@emollick.bsky.social
|
To be clear, not all the effects will be good, some will he bad and some will just be very weird. We haven’t even started to grapple with long-term impacts
| 0 |
2024-12-11T23:07:07.501Z
|
2024-12-11T23:07:07.825Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ld2uhsmty226
|
bafyreieqiqjjuts5aqjr5vfajked4yseuqhllnqc4mrzaanj23y2vzzjge
|
<start>If AI development stopped this week we would have 5-10 years of absorbing the impact of current models on education, culture, healthcare, and business.
But this week has also suggested that development is not stopping. To be clear, not all the effects will be good, some will he bad and some will just be very weird. We haven’t even started to grapple with long-term impacts
@emollick.bsky.social
|
If AI development stopped this week we would have 5-10 years of absorbing the impact of current models on education, culture, healthcare, and business.
But this week has also suggested that development is not stopping.
| 1 |
2024-12-11T23:03:36.281Z
|
2024-12-11T23:03:36.522Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ld25xyfcec2c
|
bafyreia7wxxney7rjtv3nsabvopdqvn5tlwhtuk5h7xzkido7w4amk4fue
|
<start>One warning to instructors is that the new Google Deep Research feature solves most of the issues with AI-created research assignments, at least at the undergraduate & below level
Pretty solidly well-organized and written, with accurate citations, it makes it very easy for students to skip research
@emollick.bsky.social
|
One warning to instructors is that the new Google Deep Research feature solves most of the issues with AI-created research assignments, at least at the undergraduate & below level
Pretty solidly well-organized and written, with accurate citations, it makes it very easy for students to skip research
| 0 |
2024-12-11T16:21:03.134Z
|
2024-12-11T16:21:04.030Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ld252bz5v22c
|
bafyreica7erbai7nnaeuoquebqvye4wtjesmwyjmmv7hmdr5g3fl7f55ti
|
<start>I wish they had stats on the hallucination rate. I suspect better than an undergraduate, and it is more likely to miss subtle things than to get stuff completely wrong.
@emollick.bsky.social
|
I wish they had stats on the hallucination rate. I suspect better than an undergraduate, and it is more likely to miss subtle things than to get stuff completely wrong.
| 0 |
2024-12-11T16:04:26.589Z
|
2024-12-11T16:04:27.525Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3ld24vx5go22c
|
bafyreibfceji4ryi5e4dpor3ayz4lcf5iem7e27zgdif3gpnb26fdmj26i
|
<start>The new Deep Research feature from Google feels like one of the most appropriately "Google-y" uses of AI to date, and is quite impressive.
I've had access for a bit and it does very good initial reports on almost any topic. The paywalls around academic sources puts some limits around it, though. I wish they had stats on the hallucination rate. I suspect better than an undergraduate, and it is more likely to miss subtle things than to get stuff completely wrong.
@emollick.bsky.social
|
The new Deep Research feature from Google feels like one of the most appropriately "Google-y" uses of AI to date, and is quite impressive.
I've had access for a bit and it does very good initial reports on almost any topic. The paywalls around academic sources puts some limits around it, though.
| 1 |
2024-12-11T16:02:00.972Z
|
2024-12-11T16:02:03.726Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcyo6aanc22w
|
bafyreicugtsraehiemztetrui65g22kfxnp5tljdnuhqg34wmeskskfmde
|
<start>"Claude, entertain me. i am very sophisticated. do not give me choices, just entertainment."
"i am sophisticated, not a nerd"
"more sophisticated"
"more sophisticated! I demand sophistication"
(the last line is actually pretty great) "Provide unbearable levels of sophistication"
@emollick.bsky.social
|
"Claude, entertain me. i am very sophisticated. do not give me choices, just entertainment."
"i am sophisticated, not a nerd"
"more sophisticated"
"more sophisticated! I demand sophistication"
(the last line is actually pretty great)
| 1 |
2024-12-11T02:05:33.076Z
|
2024-12-11T02:05:38.418Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcy7pawhkc2i
|
bafyreiagncv4amq2qjmlvrsiwnkovhen6qo6yguoalmarysufqo7addn5e
|
<start>OpenAI has a lot of pieces on the board right now - multimodal vision & voice; small, large, & reasoning models; image and video creation; code execution; mobile and desktop apps; web search; some agentic stuff…
Very curious when it will be glued together into A Singular Thing.
@emollick.bsky.social
|
OpenAI has a lot of pieces on the board right now - multimodal vision & voice; small, large, & reasoning models; image and video creation; code execution; mobile and desktop apps; web search; some agentic stuff…
Very curious when it will be glued together into A Singular Thing.
| 0 |
2024-12-10T21:46:38.102Z
|
2024-12-10T21:46:38.320Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcxsd2d5q22u
|
bafyreibnlw2fmd47vsjuy5lybs2gz57ybeelessg2hasjxsge55etqral4
|
<start>Benchmarks are mostly on coding and trivia. No one is benchmarking AI for its most common uses, like the fact that people are increasingly relying on AI for psychological support and advice. Also ignoring higher order uses like relying on AI for creativity, learning & strategy. Missing so much.
@emollick.bsky.social
|
Benchmarks are mostly on coding and trivia. No one is benchmarking AI for its most common uses, like the fact that people are increasingly relying on AI for psychological support and advice. Also ignoring higher order uses like relying on AI for creativity, learning & strategy. Missing so much.
| 0 |
2024-12-10T17:47:07.793Z
|
2024-12-10T17:47:12.617Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcxscydhnc2u
|
bafyreiawaf5wsi554angs2qlfyayuoghutgtlevpwpystpmdupmadjkmt4
|
<start>AI benchmarks falling into the McNamara Fallacy:
Step 1: Measure what can be easily measured
Step 2: Disregard that which cannot be measured easily
Step 3: Presume that which cannot be measured easily isn’t important
Step 4: Say that which can’t be easily measured doesn’t exist Benchmarks are mostly on coding and trivia. No one is benchmarking AI for its most common uses, like the fact that people are increasingly relying on AI for psychological support and advice. Also ignoring higher order uses like relying on AI for creativity, learning & strategy. Missing so much.
@emollick.bsky.social
|
AI benchmarks falling into the McNamara Fallacy:
Step 1: Measure what can be easily measured
Step 2: Disregard that which cannot be measured easily
Step 3: Presume that which cannot be measured easily isn’t important
Step 4: Say that which can’t be easily measured doesn’t exist
| 1 |
2024-12-10T17:47:07.792Z
|
2024-12-10T17:47:10.825Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcxixvzimk2l
|
bafyreiejefqvkdgak7gicgq7wfjfp6odwpeuvxgalqcc66pummhe5lrb3y
|
<start>Is there a canonical list with annotations to use for few-shot?
@emollick.bsky.social
|
Is there a canonical list with annotations to use for few-shot?
| 0 |
2024-12-10T14:59:52.595Z
|
2024-12-10T14:59:55.323Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcxiwzlbo22l
|
bafyreig5psiar3lmrex5b7ov2b6u4rt3oe65tvnf3xgusbr2tna6j3h4te
|
<start>Sometimes our anthropocentric assumptions about how intelligence "should" work (like using language for reasoning) may be holding AI back. Letting AI reason in its own native "language" in latent space could unlock new capabilities, improving reasoning over Chain of Thought. arxiv.org/pdf/2412.06769
@emollick.bsky.social
|
Sometimes our anthropocentric assumptions about how intelligence "should" work (like using language for reasoning) may be holding AI back. Letting AI reason in its own native "language" in latent space could unlock new capabilities, improving reasoning over Chain of Thought. arxiv.org/pdf/2412.06769
| 0 |
2024-12-10T14:59:22.764Z
|
2024-12-10T14:59:26.824Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcwgcnvggs2l
|
bafyreig6pfjmbgljiw6alsnk2x4ladmpwqwqcgi7mhsojxepvtncdbiajy
|
<start>When you upload a Sora video to LinkedIn you can hover over an icon and see this, as a result of automated watermarking from OpenAI.
Of course, they are pretty easy to remove, but this seems like a reasonable initial step to note AI content. Wonder if Blue Sky will start doing this too
@emollick.bsky.social
|
When you upload a Sora video to LinkedIn you can hover over an icon and see this, as a result of automated watermarking from OpenAI.
Of course, they are pretty easy to remove, but this seems like a reasonable initial step to note AI content. Wonder if Blue Sky will start doing this too
| 0 |
2024-12-10T04:39:32.208Z
|
2024-12-10T04:39:34.423Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcw5k3u4rc2r
|
bafyreighjyfukh6utxzgs7terzmydsnaxf7ivceni2hbpfsafnakghldve
|
<start>To be clear, these aren’t earth-shattering discoveries. You can’t ask for “a cure for cancer” or “a new form of physics” and get anything real. These are narrow problems in fields like economics and computer science, and require experts to identify and frame them.
@emollick.bsky.social
|
To be clear, these aren’t earth-shattering discoveries. You can’t ask for “a cure for cancer” or “a new form of physics” and get anything real. These are narrow problems in fields like economics and computer science, and require experts to identify and frame them.
| 0 |
2024-12-10T02:02:38.050Z
|
2024-12-10T02:02:38.320Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcw4k2c7w22l
|
bafyreihwsnaxni5tpcroqunavu32djagn46j5nlpowmnrpc2e7ilezsyem
|
<start>I have now had multiple instances where professors have told me that they are pretty sure that o1 found something novel in their field, but that the results are technical, non-obvious and complex enough that they can't be sure without considerable checking. An interesting problem. To be clear, these aren’t earth-shattering discoveries. You can’t ask for “a cure for cancer” or “a new form of physics” and get anything real. These are narrow problems in fields like economics and computer science, and require experts to identify and frame them.
@emollick.bsky.social
|
I have now had multiple instances where professors have told me that they are pretty sure that o1 found something novel in their field, but that the results are technical, non-obvious and complex enough that they can't be sure without considerable checking. An interesting problem.
| 1 |
2024-12-10T01:44:42.673Z
|
2024-12-10T01:44:44.030Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcw2rhcmqk2l
|
bafyreihy4btv3h5lfi4js46pe4nekqsjvtbalzcwmr3mjbehnanydfz4le
|
<start>Here's the second video generated in the pair. As you can see instruction following is only okay, and there is a tendency towards slow-motion action that you see everywhere.
I would also like to see some of the same testing we saw about biases in image models for new video models.
@emollick.bsky.social
|
Here's the second video generated in the pair. As you can see instruction following is only okay, and there is a tendency towards slow-motion action that you see everywhere.
I would also like to see some of the same testing we saw about biases in image models for new video models.
| 0 |
2024-12-10T01:13:03.715Z
|
2024-12-10T01:13:05.716Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcw2netk7c2l
|
bafyreihpmtpebisxxyda2tbhiscirxkri3ccvsahcvkxvgiqnqd7u4dp6e
|
<start>An advantage of Sora is that it has high character consistency. I tried to replicate the demo video of the female viking warrior with: "female knight, face splattered in mud, swings a sword at a skeleton as the camera zooms in on her face"
Note the complex mud patterns hold across 2 shots, but... Here's the second video generated in the pair. As you can see instruction following is only okay, and there is a tendency towards slow-motion action that you see everywhere.
I would also like to see some of the same testing we saw about biases in image models for new video models.
@emollick.bsky.social
|
An advantage of Sora is that it has high character consistency. I tried to replicate the demo video of the female viking warrior with: "female knight, face splattered in mud, swings a sword at a skeleton as the camera zooms in on her face"
Note the complex mud patterns hold across 2 shots, but...
| 1 |
2024-12-10T01:10:46.906Z
|
2024-12-10T01:10:48.630Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcvmzox5zc2l
|
bafyreibquziw4zhd54acquzl4ltznthqpe7ucfvvny7kla3gq6dugi2xvm
|
<start>"golden retriever running through a brutalist art gallery with modern art on the walls as a single red balloon floats overhead" in Sora, Kling (the best Chinese model), Runway (best American model). Best of 2 videos
This is a hard prompt, so look at consistency, reflections, etc
@emollick.bsky.social
|
"golden retriever running through a brutalist art gallery with modern art on the walls as a single red balloon floats overhead" in Sora, Kling (the best Chinese model), Runway (best American model). Best of 2 videos
This is a hard prompt, so look at consistency, reflections, etc
| 0 |
2024-12-09T21:07:07.777Z
|
2024-12-09T21:07:10.264Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcuyfcer6k2u
|
bafyreif4z5aze3qu7ykbaydt3tn7ymh72n2hxp467r7ergez754ys3tg3u
|
<start>I wrote a list of 15 types of work where AI is particularly helpful, and also 5 areas where it can be a trap.
AI is still best, at least for now, as a co-intelligence working with humans. Knowing when to use it, and when to avoid it, is a useful skill. www.oneusefulthing.org/p/15-times-t...
@emollick.bsky.social
|
I wrote a list of 15 types of work where AI is particularly helpful, and also 5 areas where it can be a trap.
AI is still best, at least for now, as a co-intelligence working with humans. Knowing when to use it, and when to avoid it, is a useful skill. www.oneusefulthing.org/p/15-times-t...
| 0 |
2024-12-09T14:57:48.666Z
|
2024-12-09T14:57:50.969Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lctpv2kp422l
|
bafyreigk33styofs3x64jauu6vvnjahaqgf3d2b4tnvia2x2bu4p2xeuty
|
<start>"Chekov's Gun, but for sociology - is that anything?"
Claude does a good job explaining, but that o1 answer is pretty impressive, connecting the concept to a variety of theories in an interesting way.
@emollick.bsky.social
|
"Chekov's Gun, but for sociology - is that anything?"
Claude does a good job explaining, but that o1 answer is pretty impressive, connecting the concept to a variety of theories in an interesting way.
| 0 |
2024-12-09T02:52:53.925Z
|
2024-12-09T02:52:56.672Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lct25vlqgc22
|
bafyreibch6y35n2py3wqfx5lte45p334qs3spzl6wy6tuglf4lx7xpo4eq
|
<start>It is kind of hard to do “serious” o1 demos that impress people who aren’t experts because they don’t have baseline expectations for expert human work in a field.
That is why I tend to share examples that are more accessible. It doesn’t mean o1 can’t do real work (o1 isn’t flawless, though)
@emollick.bsky.social
|
It is kind of hard to do “serious” o1 demos that impress people who aren’t experts because they don’t have baseline expectations for expert human work in a field.
That is why I tend to share examples that are more accessible. It doesn’t mean o1 can’t do real work (o1 isn’t flawless, though)
| 0 |
2024-12-08T20:24:08.389Z
|
2024-12-08T20:24:08.620Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcso6glpk225
|
bafyreibpyskz73w6gwruaang4pvjkau6hja54ugly4y2ao32pjuhnowl3m
|
<start>The image output of all of the current models is even more disconnected from the model itself, the LLM is just prompting a seperate image generator (though that will change very soon with full multimodal).
@emollick.bsky.social
|
The image output of all of the current models is even more disconnected from the model itself, the LLM is just prompting a seperate image generator (though that will change very soon with full multimodal).
| 0 |
2024-12-08T16:49:41.313Z
|
2024-12-08T16:49:41.715Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcso237b6225
|
bafyreigeo5aipxolygx3dzkvcat3r75i3dbdnrjerv5bzyjyaqneodmfme
|
<start>The vision capabilities of AI models often lag their other abilities. Great at text, often bad at intricate details in hard-to-predict ways. If you give the AI a picture of a geometry problem and it fails, it may just be the vision model, for example. The image output of all of the current models is even more disconnected from the model itself, the LLM is just prompting a seperate image generator (though that will change very soon with full multimodal).
@emollick.bsky.social
|
The vision capabilities of AI models often lag their other abilities. Great at text, often bad at intricate details in hard-to-predict ways. If you give the AI a picture of a geometry problem and it fails, it may just be the vision model, for example.
| 1 |
2024-12-08T16:47:15.152Z
|
2024-12-08T16:47:15.614Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcrmwvuwhs2h
|
bafyreiccjmpfgfhkcm7aismgwnoxeyoenkhbomcr7uptnnqjvzlmdzbxhe
|
<start>Sharks are older than the rings of Saturn.
This paper finds that the rings are no older than 400M years. Sharks date back to at least the Late Ordovician Period, 450M years ago.
@emollick.bsky.social
|
Sharks are older than the rings of Saturn.
This paper finds that the rings are no older than 400M years. Sharks date back to at least the Late Ordovician Period, 450M years ago.
| 0 |
2024-12-08T06:54:55.425Z
|
2024-12-08T06:54:57.016Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcrl6sxghs2h
|
bafyreibfojhpj5siprf3rmas7gkpx6ma4lzx5l43fnilkkgwehblnhky3a
|
<start>Samples from some experts It is kind of hard to do “serious” o1 demos that impress people who aren’t experts because they don’t have baseline expectations for expert human work in a field.
That is why I tend to share examples that are more accessible. It doesn’t mean o1 can’t do real work (o1 isn’t flawless, though)
@emollick.bsky.social
|
Samples from some experts
| 1 |
2024-12-08T06:23:33.314Z
|
2024-12-08T06:23:35.015Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcrkwboogs2h
|
bafyreibgxo64fnf7uoixto4bm4pm5whyb4puisi6iv4fzomwjtughlx57m
|
<start>Agree on code interpreter missing being a shame, but you should ask it to output code to run. On papers, you would be surprised at how well it works copying text from a PDF with screenshots of key pages of formulas. Worth trying.
@emollick.bsky.social
|
Agree on code interpreter missing being a shame, but you should ask it to output code to run. On papers, you would be surprised at how well it works copying text from a PDF with screenshots of key pages of formulas. Worth trying.
| 0 |
2024-12-08T06:18:46.768Z
|
2024-12-08T06:18:47.013Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcrizqra6s2e
|
bafyreiflpz7ydn7m27lyllkzsogiu2kvl3wgdz5blqdz6dsacpsu5bshcy
|
<start>We need a wider network of experts to verify the output of cutting-edge models. I am impressed by o1 in fields I know well, it cracks HBS cases & analyzes papers in economic sociology
But legal work? Diagnosis? Operations research? Electrical engineering? Logistics? Paleontology? Literary analysis? Samples from some experts
@emollick.bsky.social
|
We need a wider network of experts to verify the output of cutting-edge models. I am impressed by o1 in fields I know well, it cracks HBS cases & analyzes papers in economic sociology
But legal work? Diagnosis? Operations research? Electrical engineering? Logistics? Paleontology? Literary analysis?
| 1 |
2024-12-08T05:44:55.759Z
|
2024-12-08T05:44:56.511Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcrfcljx322e
|
bafyreia4nh2jzd25wx2yf6ytesdg75pnz5wnq7rxs75lrvg7frni2vxmim
|
<start>Lots of people experimenting and deploying, few talking publicly.
@emollick.bsky.social
|
Lots of people experimenting and deploying, few talking publicly.
| 0 |
2024-12-08T04:38:17.295Z
|
2024-12-08T04:38:18.314Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcrc464f222e
|
bafyreidgbhfzh4lz4krgw7v2pt6auuk35ruphdi5jdyj6ybvc34u6mfbba
|
<start>Can't find anything like it by Googling, the chain of thought started in other places, and the fact that it modifies it on command suggests yes
@emollick.bsky.social
|
Can't find anything like it by Googling, the chain of thought started in other places, and the fact that it modifies it on command suggests yes
| 0 |
2024-12-08T03:41:00.673Z
|
2024-12-08T03:41:01.320Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcrbzt2nl22e
|
bafyreigjdqbw23b3p6hf6rvo56yguh2ec4rljctddzzf6g332gdd73rueu
|
<start>👀"that's pretty darn good, but you don't need the letters in order and maybe the silent e is a problem? can you do some minor revisions?"
@emollick.bsky.social
|
👀"that's pretty darn good, but you don't need the letters in order and maybe the silent e is a problem? can you do some minor revisions?"
| 0 |
2024-12-08T03:39:41.970Z
|
2024-12-08T03:39:44.411Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcrbe5cp522e
|
bafyreihwozmpsegtx6v2dlilffiawp4z4saofce5oawb7vcmdxndvg2yk4
|
<start>Here are the hints it gave and the solution.
(There is one unnecessary step in the puzzle, by the way, but a zero shot coherent word puzzle is pretty neat) 👀"that's pretty darn good, but you don't need the letters in order and maybe the silent e is a problem? can you do some minor revisions?"
@emollick.bsky.social
|
Here are the hints it gave and the solution.
(There is one unnecessary step in the puzzle, by the way, but a zero shot coherent word puzzle is pretty neat)
| 1 |
2024-12-08T03:27:33.847Z
|
2024-12-08T03:27:37.183Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcrbe4nzys2e
|
bafyreiaxl4yttlkxghmxlxjlcrcccoz4gdcz7lemqh4lrxugrv32qqggai
|
<start>I thought puzzle design beyond current AI, so this is the first time I have seen a model actually create a solvable (if not perfect) and interesting puzzle.
"o1, create a d&d puzzle that is non-trivial, but solvable. it should not involve elements or statues, be creative."
(Solution in thread) Here are the hints it gave and the solution.
(There is one unnecessary step in the puzzle, by the way, but a zero shot coherent word puzzle is pretty neat)
@emollick.bsky.social
|
I thought puzzle design beyond current AI, so this is the first time I have seen a model actually create a solvable (if not perfect) and interesting puzzle.
"o1, create a d&d puzzle that is non-trivial, but solvable. it should not involve elements or statues, be creative."
(Solution in thread)
| 1 |
2024-12-08T03:27:33.846Z
|
2024-12-08T03:27:36.814Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcr2brv6ic25
|
bafyreiggik2sm7vdvficoqtjb3lxsf3xtd6yuypsnsgoyea4jw2nz6yj5u
|
<start>I suspect that linking back to twitter is not done here, but this is a fascinating look into a case of how LLMs end up doing weird things that is pretty illuminating.
Specifically, why does an LLM constrained to only be able to use words from the Bible keep saying “ouches”? x.com/voooooogel/s...
@emollick.bsky.social
|
I suspect that linking back to twitter is not done here, but this is a fascinating look into a case of how LLMs end up doing weird things that is pretty illuminating.
Specifically, why does an LLM constrained to only be able to use words from the Bible keep saying “ouches”? x.com/voooooogel/s...
| 0 |
2024-12-08T01:20:59.245Z
|
2024-12-08T01:20:59.528Z
|
emollick.bsky.social
|
did:plc:flxq4uyjfotciovpw3x3fxnu
|
at://did:plc:flxq4uyjfotciovpw3x3fxnu/app.bsky.feed.post/3lcqt2hgxx22a
|
bafyreid3m2amfzxo3olohvdmj4e4uyonbt56dbvbos24eu3ps4yd7gtzwu
|
<start>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
|
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.
| 0 |
2024-12-07T23:11:37.221Z
|
2024-12-07T23:11:37.421Z
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.