The Godfather in Conversation: Why Geoffrey Hinton is worried about the future of AI

Geoffrey Hinton, known to many as the “Godfather of AI,” recently made headlines around the world after leaving his job at Google to speak more freely about the risks posed by unchecked development of artificial intelligence, including popular tools like ChatGPT and Google’s PaLM.

Why does he believe digital intelligence could hold an advantage over biological intelligence? How did he suddenly arrive at this conclusion after a lifetime of work in the field? Most importantly, what – if anything – can be done to safeguard the future of humanity? The University of Toronto University Professor Emeritus addresses these questions and more in The Godfather in Conversation.

00:00 Intro
01:03 Digital intelligence
02:27 Biological intelligence
03:47 Why worry?
04:39 Machine learning
07:07 Neural Nets
13:22 Neural nets and language
17:18 Challenges
18:49 Breakthrough moment
20:41 AlexNet
24:35 Pace of Innovation
26:04 ChatGPT
27:46 Public Reaction
29:49 Benefits for society
33:25 Pace of innovation
35:48 Sudden realization
37:13 Role of government
40:08 Big tech
42:32 Advice to researchers
43:50 Understanding risk
45:20 What’s next?

[Douchebagus]

This man is a voice of reason, honesty, and inspiration in a world of chaos. I am grateful that this man existed and was able to materialize his gifts, even if they may destroy us all.

[olachalikova]

I'll be honest, I started listening in the middle of the night hoping it would put me back to sleep as complex things sometimes help my brain shut off. But Mr Hinton seems to be AMAZING at explaining complex things in a easily digestible way. On top of that I was impressed by his humility and trust in collective humanity. This video should totally be added to school curriculums.

[sussechandrasekaran]

“If you can't explain something in simple terms, you don't understand it.” Richard Feynman . Hinton in this video does this quote justice !

[loonpohchuah]

Hinton explain things clearly and is honest with himself!

[carl]

These explanations of neural nets are really brilliant but his humility and honesty really really stood out here too. Want to consume me a load more of Hinton's work

[Learna_Hydralis]

I love Hinton, I think he is honest character and driven by great stuff like curiosity and desire to understand and not motivated be money, power.. etc.

[inteligenciamilgrau]

27:21 "The idea that it's just sort of predicting the next word and using statistics - there's a sense in which that's true, but it's not the sense of statistics that most people understand. It, from the data, it figures out how to extract the meaning of the sentence and it uses the meaning of the sentence to predict the next word. It really does understand and that's quite shocking."
29:41 "the idea that it's just predicting the next word doesn't make any sense in that context. It had to understand what was going on"

[henrykuyvenhoven]

This is a must watch for anyone who wants peace, prosperity, liberty and the striving to be beautiful as a global community.

[mahdikashanipour]

Such a talk! He explained neural networks and back propagation, etc amazingly! He is right.

[nicklansbury]

This was a very insightful interview. Not only were the questions well-considered but Geoffrey Hinton answered them in a sympathetic and respectful way. Thank you for arranging this interview and for uploading it.

[SolaceEasy]

You are what you wish you could be. Geoffrey Hinton, you are a philosopher.

[dewakarprasad]

He was clearly open to hint that the risks of AI getting out of control and being Godfather of AI we should think about it seriously

[SquashFiend]

I just hope in hindsight we will not look at Geoffrey Hinton as the Oppenheimer of our times, a man who invented something extra-ordinary which he could exert no control over it's proliferation and created a new world

[squamish]

What's interesting is how everyone, even experts, were completely wrong about the timeline of AI. This sort of stuff - human-level intelligence, or artificial general intelligence - was not predicted until a median of 2050, as recently as 2017! Now, Google just published a paper predicting that artificial _super_ intelligence (far better than, like a million - pick a number - humans combined at everything) may be here by 2030.

People immediately accused Google of hype for $$$. But Hinton has no such stake in predicting this stuff, and yet he is doing so anyway, with the same timeline. As he says, he recently thought AGI was 30 to 50 years away. Then in March 2023, he thought it was less than 20 years away. Now, as of this interview in June 2023, he thinks that AGI may be here within a few years. He changed his stance completely in a matter of less than a year.

And Hinton truly has been around since the very beginning. He helped found a charity for computational neuroscience in 1967, when he was 20 years old, and computers were as large as rooms.

[TheMrCougarful]

The very idea that AGI can ever be controlled is itself dangerous. The possibility of anyone finding themselves in the position of saying, "we have this under control", means it will absolutely guaranteed escape containment. The human record for evaluating these risks, and coming up way too optimistic, is vast and compelling.

[fmjrvfi]

sincere, trustworthy, sympathetic, mixed with humor

[antigonemerlin]

I just want to say I love the video editing here. It's both subtle, and incredibly creative. Video editing is one of those places where if you do well, nobody notices, so I just want to give a big shoutout to whoever did that!

[briancase]

There's a lot of overlap with previous interviews, but that's no problem at all. I can listen to Jeffrey explain things many more times. And, he covers a bit of me ground here thanks to God questions. Thanks very much for this!

[unemployedrocketsurgeon]

He’s a genius, a lot of people are intelligent but not very wise, this guy has both in spadefuls. Let’s listen & do something to control this thing.

[trybunt]

That story about the programmer writing a new language to test GPTs understanding gave me goosebumps. I would live to hear more about that. Maybe the most compelling story ive heard, if true, i would be convinced it really does understand