top of page
Writer's pictureMal McCallion

Keeping AI friends close ...

Updated: Dec 11, 2023


In 2019 a tiny portion of the world celebrated when the three 'Godfathers of AI' - Geoff Hinton, Yann LeCun and Yoshua Bengio, won the Turing Award (and a share of $1m) for their work on AI. This is often called the Nobel Prize of tech. For those involved it was, at last, a coming of age.


Decades of hard work - often through the 'AI Winters', when everyone else decided to pack their hard drives away and go and do something more fun like crypto - had finally been recognised. Neural networks, a specific type of 'machine learning' that tries to replicate the complex patterns and overlays of our own brain's thinking, were finally starting to prove what these guys had believed as far back as the 1980's; artificial intelligence was real. (A great book on this, by the way, is 'Genius Makers' by Cade Metz - highly recommend.)


The rest of the world, of course, had no clue any of this was happening and carried on trying to think of the right hashtag combo for their 280-character Tweets.


Fast-forward five years and the Godfathers don't share the same views on the future of their 'creation' any more. Once firmly of the opinion that their creation was benign, Hinton and Bengio now think that the risks are now significant that 'bad actors' will get hold of the tech and potentially wipe us all out. LeCun, on the other hand, soothingly strokes our forehead and tells us it's all going to be alright.


Whether commercial advantage plays a part in LeCun's equanimity is a genuine question - he resides at the right-hand of Mark Zuckerberg (nudging Nick Clegg up the sofa a bit) at Meta, for whom AI is an important play after the hilariousness of the Metaverse. LeCun has been a champion of 'open source' within AI - enabling everyone to get hold of the tech so that good things can be crowdsourced from as many brainy people as possible.


Unfortunately, of course, brainy people can be bad people too.


Whoever is right - and Hinton believes in the risks so much that he's resigned from his highly-lucrative Google gig to tell us all about them - remains to be seen. But 'seen' it will be - LeCun has released Meta's foundational model LLaMA out into the world for us all to build upon.


The genie is firmly out of the bottle. Which of the Godfathers is right, we're most definitely going to find out.


Buckle up.



Made with TRUST_AI - see the Charter: https://www.modelprop.co.uk/trust-ai

5 views0 comments

Comments


bottom of page