'The godfather of AI' warns about the risks of AI

Geoffrey Hinton, computer scientist and cognitive psychologist, joins BNN Bloomberg to discuss the existential threat of AI along with his part in shaping the world of AI.

Connect with BNN Bloomberg:
--
BNN Bloomberg is Canada’s only TV service devoted exclusively to business, finance and the markets.

[djayjp]

Lol! AI is already smarter than most humans. There have been tests performed in which AI beats humans at reasoning tasks.

[eklim]

Evolution is a function of self-interest which is a function of growing intelligence, i.e. if intelligence keep growing, at certain point self-interest kicks in

[m_go]

Former Head of Security says that by 2027 AGI is achieved and then Superintelligence shortly after.

[denisblack]

Funny how AI is a scam, but layoffs are real 😅
Guess that’s how they chose to cover up for financial crisis

[dave-]

There is an old film that illustrates this perfectly- colossus the forbin project. A good watch in my opinion.

[currawong]

Smarter than us? Judging by the current state of world affairs and those who are responsible, it would not be difficult to be smarter than us.

[mahiaravaarava]

It’s always intriguing to hear insights from AI pioneers, especially when they highlight the potential risks involved. It's fascinating to think about how platforms like SmythOS are stepping in to help organizations harness AI collaboratively.

[vikasrai]

Already even creators don't understand how these chatbots work. It puts us in a 'national security' kind of situation. If you can't understand an intelligent or very intelligent being, it's a big concern.

[HaraldEngels]

To be smarter than humans doesn't need much. Just look to the status of the world, it says everything about our intelligence.

[arknewman]

I’m having trouble keeping track of all these godfathers of AI.

[malcolmwatt]

It becomes a legal issue. If AI becomes and I've heard it expressed that AI has made death threats against humans this is a legal matter. Death threats are to my knowledge against the law. If AI is to become a mass murderer then those responsible are accessories and can be prosecuted. Even the threat of mass murder is sufficient for prosecution. I would suggest that a class action suit against the AI researchers or federal charges be laid against the perpetrators with warrants and arrests. Otherwise, chaos will ensue.

[sphan]

When the Godfather speaks, we should probably listen.

[raul]

Let him repay all the millions of dollars he has earned developing artificial intelligence over the years, before becoming old enough to not be able to compete against the younger ones. Now he is dedicated to preaching moral teachings. Paradoxical.

[AAL]

AI does not have to be smarter than us. It just needs to be good enough to do what we do. It not decades away. It's already having an impact now. Eventually it may be an existential risk but what about the significant impacts in the near term? They are indirectly an existential threat now. We need to harness it use to help humans flourish and augment our abilities and to serve our goals. Not replace us.

[ChristopherBruns-oo]

3:48 So like when regimes act out and they shut off the internet to certain areas of the world as crowd control - why would this not help?

[drblitz]

All this Ai, and I’m still doing the same thing I’ve been doing past 15 years

[AIRR]

either nuclear bombs kills us or AI kills us. In the end we die

[Aggielife]

The immediate threat of anything right now are jobs. Specifically computer bases white collar jobs first and then other jobs later on. If you have a job primarily based around data entry, search, processing, accounting, etc. start looking for something else now! Don’t let it catch you off guard like those people working at Indeed that found themselves out of work looking for another job on the site they use to work at!

[KMegahertz]

Even with transformers, it still boils down to a "look up table" of sorts. You could remove maybe or or two lines of code from the LLM that randomize the next predicted word from the most probable ones and feed it in an input and get a result. Feed that same input again, and you would get the same result, over and over again. There is no intelligence going on. Feel free to correct me on this if I am wrong.

[The_AI_Adoption]

"The Level 2 of AI Adoption is arriving before AGI"

The Last AI of Humanity (New Book)