Topics
Latest
AI
Amazon
Image Credits:Bryce Durbin / TechCrunch
Apps
Biotech & Health
mood
Image Credits:Bryce Durbin / TechCrunch
Cloud Computing
Department of Commerce
Crypto
enterprisingness
EVs
Fintech
Fundraising
Gadgets
punt
Government & Policy
Hardware
Layoffs
Media & Entertainment
Meta
Microsoft
Privacy
Robotics
Security
Social
infinite
Startups
TikTok
exile
Venture
More from TechCrunch
Events
Startup Battlefield
StrictlyVC
Podcasts
Videos
Partner Content
TechCrunch Brand Studio
Crunchboard
Contact Us
AI company founders have a report for crap bold claims about the engineering ’s potential to remold field , particularly the sciences . But Thomas Wolf , Hugging Face ’s cobalt - founder and chief scientific discipline ship’s officer , has a more measured take .
In anessay print to Xon Thursday , Wolf said that he fear AI becoming “ yes - men on server ” remove a discovery in AI enquiry . He elaborated that current AI maturation epitome wo n’t buckle under AI adequate to of outside - the - box , originative problem - resolution — the kind of problem - solving that wins Nobel Prizes .
“ The main fault citizenry ordinarily make is intend [ people wish ] Newton or Einstein were just scaled - up safe pupil , that a genius comes to life when you linearly generalise a top-10 % student , ” Wolf write . “ To produce an Einstein in a data point center field , we do n’t just ask a organisation that knows all the answers , but rather one that can ask questions nobody else has thought of or dared to ask . ”
Wolf ’s assertions tolerate in contrast to those from OpenAI CEO Sam Altman , who in anessay earlier this yearsaid that “ superintelligent ” AI could “ massively accelerate scientific discovery . ” Similarly , Anthropic CEO Dario Amodei has prognosticate AI couldhelp give voice cures for most case of cancer .
Wolf ’s problem with AI today — and where he thinks the engineering is head up — is that it does n’t sire any new noesis by connecting antecedently unrelated fact . Even with most of the internet at its administration , AI as we currently sympathise it mostly fill up in the spread between what mankind already know , Wolf say .
Some AI expert , includingex - Google engineer François Chollet , have give tongue to interchangeable views , arguing that while AI might be adequate to of memorize reasoning patterns , it ’s unlikely it can generate “ new logical thinking ” base on novel situations .
Wolf opine that AI labs are build what are essentially “ very obedient students ” — not scientific revolutionary in any sensory faculty of the idiom . AI today is n’t incentivized to question and propose approximation that potentially go against its education data point , he enunciate , confine it to answer known questions .
Join us at TechCrunch Sessions: AI
Exhibit at TechCrunch Sessions: AI
“ To create an Einstein in a data center , we do n’t just need a organisation that know all the solvent , but rather one that can ask interrogation nobody else has thought of or dare to take , ” Wolf said . “ One that writes ‘ What if everyone is wrong about this ? ’ when all textbooks , expert , and common knowledge suggest otherwise . ”
Wolf thinks that the “ valuation crisis ” in AI is partly to charge for this disenchanting state of affaire . He point to benchmarks commonly used to measure AI system improvements , most of which dwell of interrogation that have cleared , obvious , and “ closed - ended ” answer .
As a result , Wolf proposes that the AI industry “ move to a measure of knowledge and logical thinking ” that ’s able-bodied to sort out whether AI can take “ bold counterfactual approaching , ” make oecumenical proposition based on “ petite hints , ” and ask “ non - obvious questions ” that run to “ new research path . ”
The whoremaster will be figuring out what this measuring rod look like , Wolf take . But he think that it could be well worth the effort .
“ [ T]he most important face of skill [ is ] the skill to ask the right doubtfulness and to challenge even what one has see , ” Wolf read . “ We do n’t call for an A+ [ AI ] student who can answer every question with universal knowledge . We need a B student who sees and motion what everyone else missed . ”
