2017年に出されたAI予測の論文が面白い 2023: 小説の内容を基にグループ化、インターネット上の情報を検索して回答、スピーチの文字起こし 2024: ロボットアームで任意のレゴを組み立てる、すべてのAtariゲームを人間より上手くプレイ、電話オペレーター https://t.co/2nA9OWUqPi
RT @tamaybes: Many AI experts clearly don't think this is science fiction, and AI labs are spending hundreds of billions to make it happen.…
Sure but AI researchers have incentives to say that - their funding relies on persuading financial backers that they will keep making big brealthroughs quickly. LLMs are impressive and can improve research productivity but AGI is far away
RT @GordonBrianR: But also, while many people in AI coming from CS are pretty optimistic about the potential for near-term AGI/ASI arising…
But also, while many people in AI coming from CS are pretty optimistic about the potential for near-term AGI/ASI arising out of the current stack, many from cogsci and neuro are much more cautious. So there is no consensus of experts on this at all.
Pessimistic induction?
RT @tamaybes: Many AI experts clearly don't think this is science fiction, and AI labs are spending hundreds of billions to make it happen.…
Many AI experts clearly don't think this is science fiction, and AI labs are spending hundreds of billions to make it happen. Why do economists defer so little to AI experts about the topic of what AI can or can't do? https://t.co/WWIdocrNRB https://t.co/j
@tomazfernandes_ @sseraphini Segunda imagem em: https://t.co/xpeSHoFBw6
@CornelWest And stay far away from yang his plan was shit. The tax structure also needs updated as excess is what needs to be redistributed, we do not want to switch to consumption taxes. https://t.co/dkLjgiBm0b
This comparison on 2022 and 2023 shows that, but if you go even further back to the original analysis in 2016 it gets more fun. https://t.co/5Dh44Jwdjz https://t.co/aYOO3BSPq4
@ScandyAndalous @RylanSchaeffer We actually also have 2016, with mostly identical questions (https://t.co/I1dvHZtzDi)
Always fun to look at AI predictions from a few years ago. Probably one of the most famous ones: https://t.co/7jrzx5hf9C
Always fun to look AI predictions from a few years ago. Probably one of the most famous ones: https://t.co/7jrzx5hf9C
@guilhaweb “artificial intelligence will automate all human tasks in the next 45 years and all human jobs in the next 120 years” 🤣🤣🤣🤣🤣🤣🤣🤣🤣 https://t.co/I0mpK7ktz9
AI milestones based on predictions of Oxford Uni researchers https://t.co/i05BcrPqor https://t.co/vDTa3EQGnJ
Paper: https://t.co/rmlEbqO4Ci
@WojciechW13 @Gabriel02255244 Check out the abstract! https://t.co/WtGAwHxARv
@Gabriel02255244 samo badanie: https://t.co/ykgXisPe00 Oxford University i Yale University
@stoarca @foomagemindset Check out the abstract! https://t.co/WtGAwHxARv
@foomagemindset Here is a 2017 paper about a 2015 survey of AI experts: https://t.co/wQ1cQZ7l7G You can see that almost all of the predictions that have been fulfilled so far have happened near the lower end of the bounds, not the median. The only except
@HannesThurnherr @forkinggizmo @paulg @FrMersenne Check out the abstract! https://t.co/WtGAwHxARv
@forkinggizmo @paulg @FrMersenne this doesn't have to sum up the topic. But this is Geoffrey Hinton who was instrumental in the development of modern deep learning. Until recently, he wasn't very concerned. Now he is. Here is a study with a representative
@HannesThurnherr @LeonDerczynski Check out the abstract! https://t.co/WtGAwHxARv
@LeonDerczynski Well, it has strong parts. Again the one where it lists the dangers that have come to pass. Though listing the CO2 output is laughable. And my tweets aren't the only things countering it by far! Again: https://t.co/q8PU7Owldl
@HannesThurnherr @sibinmohan @TaliaRinger @LeonDerczynski Check out the abstract! https://t.co/WtGAwHy8H3
@LeonDerczynski @sibinmohan @TaliaRinger where did i do that? https://t.co/7TD3Gy084T
@sibinmohan @TaliaRinger @LeonDerczynski Guys, please. The whole point i'm making is that none of us get to take away someone's staus as a scientist for having a different opinion. So, I think they are a serious scientists But so are the hundreds of high-
@HannesThurnherr @TaliaRinger @LeonDerczynski Check out the abstract! https://t.co/WtGAwHxARv
@TaliaRinger @LeonDerczynski I'm talking about actual AI researchers. Are you? https://t.co/q8PU7Owldl
@needstoo1 @__eyja @kestrel1hawk2 Check out the abstract! https://t.co/WtGAwHxARv
@__eyja @kestrel1hawk2 70% of AI researchers think AGI might pose a serious problem. https://t.co/30CXZiPZpT
@DoniyaConseil …à quel moment l’IA serait capable de remplacer les humains sur des postes considérés aujourd’hui comme "humains" uniquement? https://t.co/0Edk28URgj
Les machines surpasseront les humains dans un grand nombre de tâches d'ici à 2062 d'après cette étude #IA https://t.co/75KDMcnUnb
@RomanianLiberal @ChimpIRL @PogFishPoster Check out the abstract! https://t.co/WtGAwHxARv
@ChimpIRL @PogFishPoster AGI has everything to do with out performing humans lol. As for my claim, you can check surveys of AI researchers on this (eg: https://t.co/BSsrROkxnD) or AI development prediction models like the bio anchor model (https://t.co/Zs
@leblaison Check out the abstract! https://t.co/WtGAwHxARv
Les machines surpasseront les humains dans un grand nombre de tâches d'ici à 2062 d'après cette étude #IA https://t.co/989OrzCeEt
@LuckmannDr Check out the abstract! https://t.co/WtGAwHxARv
RT @boibombeiro: @cartermariano @import_robs Timeline of Median Estimates (with 50% intervals) for AI Achieving Human Performance. Number o…
@cartermariano @import_robs Timeline of Median Estimates (with 50% intervals) for AI Achieving Human Performance. Number of years from 2016. Viewpoint: When Will AI Exceed Human Performance? Evidence from AI Experts https://t.co/KQQUh8PTtb https://t.co/U
@BishopPair @DenkenAnalysis @RichardMCNgo @robertskmiles Check out the abstract! https://t.co/WtGAwHxARv
@DenkenAnalysis @RichardMCNgo @robertskmiles Paper is this one: https://t.co/NkP6M28yHP Apologies - I remembered the median estimate as being 2045, but it was 45 years from 2016, so 2061. I knew there was a 45 involved! So longer timeline than I suggested
@XiXiDu Check out the abstract! https://t.co/WtGAwHxARv
For reference: https://t.co/G46iyTVzIE Cf. In 2016, surveyed AI experts predicted that "Perform as well as the best human entrants in the Putnam competition" would take 34 years to achieve, in 2050: https://t.co/dXKEbtA6pV
RT @emollick: The AI paper is fascinating, less than 6 years old and already out of date. Many of these milestones are already passed. You…
RT @emollick: The AI paper is fascinating, less than 6 years old and already out of date. Many of these milestones are already passed. You…
RT @emollick: The AI paper is fascinating, less than 6 years old and already out of date. Many of these milestones are already passed. You…
@emollick Check out the abstract! https://t.co/WtGAwHfZsV
The AI paper is fascinating, less than 6 years old and already out of date. Many of these milestones are already passed. You can read the paper here: https://t.co/hXRDLEYI4G https://t.co/3C6R6jtF9f
https://t.co/boWIRhbOe0 2017年の大規模な人間レベルの人工知能実現時期や労働代替時期などの調査。 2036年までにこの種のAIが出現する確率は〜20%、2060年までに〜50%、2100年までに〜70% オックスフォード、人類未来研究所調査。
@RomanianLibs @LFrantzolas Check out the abstract! https://t.co/WtGAwHy8H3
@LFrantzolas Yeah, here is one study on the opinions of AI researchers on the development of AI abilities in the future. "Researchers believe there is a 50% chance of AI outperforming humans in all tasks in 45 years" https://t.co/BSsrROkxnD
@SinthePosting @PunishedBernake @RomanianLibs @TristamPratori1 @chflorelius Check out the abstract! https://t.co/WtGAwHxARv
@PunishedBernake @RomanianLibs @TristamPratori1 @chflorelius Cool. Here it is. Now answer the question I asked you. https://t.co/ntu4H7ctj0 https://t.co/gM7gsjnIKG https://t.co/ReVcgN5nX8
@bitesizedea Check out the abstract! https://t.co/WtGAwHxARv
Grace et. al. (in 2017, so apply some amount of caution about the actual numbers) asked AI experts to predict "when unaided machines can accomplish every task better and more cheaply than human workers." (26/?) https://t.co/7w7Wr8i5rP
@MythicCDXX Check out the abstract! https://t.co/WtGAwHxARv
When Will AI Exceed Human Performance? Evidence from AI Experts https://t.co/dkLjgijKBB
@SVVAYED Check out the abstract! https://t.co/WtGAwHxARv
2016 paper on AI When Will AI Exceed Human Performance? Evidence from AI Experts https://t.co/jatRnBVYIl
@yaronhadad Check out the abstract! https://t.co/WtGAwHxARv
This paper is so freaky https://t.co/yUrcbd5pS0
(3) There are few expert surveys that ask about transformative AI. Probably the only one is Grace et al. in 2016 (https://t.co/fQS80dCVoz). But responses varied depending on the question framing, with a median timeline of 2138 for full automation of labor.
@elonmusk A study of 352 researchers in the field of AI used a strict definition for general AI. The aggregate forecast gave a 50% chance of general AI occurring within 45 years and a 10% chance of it occurring by 2027. We'll see... https://t.co/UDIYKWB3sW
@fatierdogan @mehmetcagdasi Teknolojinin capital intensive sektörlerde istihdamı azaltmaması sizce de biraz gayrı-sezgisel değil mi? Biraz açar mısınız? Bu arada şu makalede hangi işler ne zaman elden gidecek'e dair güzel bir "AI yol haritası" var: https
@OldSpiceSmells @GF46949252 @AndressTheGiant @scrodumplings @thetimes Check out the abstract! https://t.co/WtGAwHxARv
[1705.08807] When Will AI Exceed Human Performance? Evidence from AI Experts --> https://t.co/zbJgux9iPB // Papers, Tech, AI
@David_Gunkel @SvenNyholm The question about timelines is a useful one. It's possible that these problems that we intuitively want to categorize as medium or long-term could occur sooner. For example, Grace et al., 2018 expert survey comes to mind: https:/
@JgaltTweets @StefanFSchubert @LinchZhang @davidmanheim @peterwildeford @CharlesD353 Check out the abstract! https://t.co/WtGAwHxARv
@StefanFSchubert @LinchZhang @davidmanheim @peterwildeford @CharlesD353 Not sure if there are any periodic expert surveys on this sort of thing; I do recall at least one survey of technical experts that included attaching probabilities to AI achieving cert
48% of AI researchers think that research on minimising the risks of AI to society should be prioritised more than it is today
AI researchers predict how many years until they believe AI will exceed human performance, across a range of abilities https://t.co/hKRGOBKoJL https://t.co/sr4AfN4mYQ
@delaaxe @Aella_Girl This isn't true, unless you think a 5% chance of human extinction is "no risk at all" https://t.co/tVUCSSFvqw
@nlpnoah If experts agreed, or we only needed to worry about one kind of error (overestimating AI), then that’d be right. But in fact views are all over and people routinely fail to anticipate developments with big societal implications months let alone ye
@vipersgratitude @ghostpants7 @Sci_Phile Check out the abstract! https://t.co/WtGAwHxARv
When Will AI Exceed Human Performance? Evidence from AI Experts https://t.co/YyFPyWslkp
@kpd_musing Check out the abstract! https://t.co/WtGAwHxARv
Had also meant to add this: https://t.co/brqLaPgNcr
[1705.08807] When Will AI Exceed Human Performance ? Evidence from AI Experts https://t.co/pLCh8uOcJU H/T: @bobehayes
“Researchers predict AI will outperform humans in many activities in the next 10 years, such as translating languages (by 2024), writing high-school essays (2026), driving a truck (2027), working in retail (2031)” @Cornell https://t.co/ZjvaMh512V #Arti
RT @future_of_AI: When Will AI Exceed Human Performance? Evidence from AI Experts https://t.co/04mA7U9pF8 #AI #Research via @Miles_Brundage
When Will AI Exceed Human Performance? Evidence from AI Experts https://t.co/04mA7U9pF8 #AI #Research via @Miles_Brundage
@IAmSamFin Yeah, I am not sure of the best way to make this rigorous but e.g. one way is looking at surveys covering a fairly wide range of things: https://t.co/VVtluIHBQL most of the forecasts were far enough out to not yet be tested, though already some
@ykilcher And also note that all of these arguments were “available” to people in advance—and yet experts thought they were hard anyway so even if you don’t update your views of AI, you should re: experts. See surveys such as https://t.co/VVtluIq0sb
@eprado_melo Check out the abstract! https://t.co/WtGAwHxARv
@herodote63 Check out the abstract! https://t.co/WtGAwHxARv
There is everything summarized: probability of HLRF* 100% Don't trust any of them to make any prediction at all. Their input data is only bias and no signal!!! (*High Level Researcher's Foolishness) https://t.co/MlrbQLXrNx
@vipersgratitude @akalamusic Check out the abstract! https://t.co/WtGAwHxARv
@leyla_bezgin Check out the abstract! https://t.co/WtGAwHxARv
RT @EstulinDaniel: Excellent report on AI https://t.co/jyYuntvSuc
Un estudio revela de qué será capaz la IA de aquí a unos años. Descubre qué trabajos que creías únicamente humanos ya no lo serán: https://t.co/hfh0O8GxW5 @KatjaGrace @JohnSalvatier #InteligenciaArtificial #Innovación
“La IA superará a los seres humanos en muchas actividades como traducir idiomas (para 2024), escribir ensayos de secundaria (2026), conducir un camión (2027), trabajar en el comercio minorista (2031), escribir un libro superventas (2049) y trabajar como c
RT @EstulinDaniel: Excellent report on AI https://t.co/jyYuntvSuc
RT @EstulinDaniel: Excellent report on AI https://t.co/jyYuntvSuc
Algo que vale el tiempo checar https://t.co/5obKCzrIpg
RT @EstulinDaniel: Excellent report on AI https://t.co/jyYuntvSuc