Can we build AI without losing control over it? | Sam Harris

Scared of superintelligent AI? You should be, says neuroscientist and philosopher Sam Harris -- and not just in some theoretical way. We're going to build superhuman machines, says Harris, but we haven't yet grappled with the problems associated with creating something that may treat us the way we treat ants.

TEDTalks is a daily video podcast of the best talks and performances from the TED Conference, where the world's leading thinkers and doers give the talk of their lives in 18 minutes (or less). Look for talks on Technology, Entertainment and Design -- plus science, business, global issues, the arts and much more.

[AitoNitram]

Sam Harris: "A global pendemic"
2020: "Hey that's me!"

[CiruPlays]

questions i gained from watching this ted talk "did ants invent humans?"

[DavidGarcia-pbzn]

I can't wait for 50 years from now; the top comment saying, "Who's here after the machine invasion? 🤔" lol

[MatiasChampi]

"To be 6 months ahead of the competition its to be 500.000 years ahead at a minimum..." recent news: Tech pioneers call for six-month pause of "out-of-control" AI development

[mordinvan]

Anyone ever wonder if the evolution of A.I. is the great filter?

[bakarenibsheut]

"A global pandemic?"
You may just have hit the nail on the head.

[freeguy]

This guy should be on the Manhatten Project version of the Board of AI Protection when its created. He has good foresight. We need to get guys like him in a room with others like him to protect us from AI getting outta control

[squamish]

"It's 50 years away" is an even worse argument when you realize it's now 45 years away and there have been a slew of breakthroughs even in a year railroaded by a global pandemic, and that the pandemic actually _accelerated_ the development of AI in some respects by increasing the already sky-high demand for it.

[minagamil]

This seems to be inevitable especially when the AI recommends this video for you ! :"3

[Kingsizeshrimp]

Can't wait till my if-else statements become self-aware...

[doruktopcu]

I thought this was a talk made 2 weeks ago.

[domitron]

Given what is going on right now with ChatGBT and other AI systems, I think Sam Harris's message here of six years ago is particularly prescient. It looks like it is happening even faster than any of us knew it would. I certainly did not believe I would see a system like ChatGBT within six years of this talk. I would have assumed more like 30 or more even. Well, we are at the threshold of the singularity now, and no one has a clue what is really going to happen.

[Reignor]

Man: Is there a god?
AI: There is now.

[matthewbateman]

The scariest part is that this is not the opening scene of a movie. This is real-life conversation happening in the real world, present day -- and given by one of the world's sharpest, most respected minds.

[amandas]

Watching this now in the Midjourney/Chat-GPT era, and yeeeah. The LLMs are especially eerie in the context of this talk because of the "black box" issue. Put simply, while the engineers and computer scientists in charge of this stuff can tell you why LLMs do *most* of the things they do and explain *most* of the answers they give, at a certain point, there's a "black box" of activity/logic/whatever happening that they can't explain. Right now, that's still in the "huh, how interesting" and "I bet they'll figure it out soon enough" stage, but given the speed at which AI and LLMs in particular are developing, I have a feeling that black box problem is going to get way worse, not better.

[willydiego]

"Powered by sunlight"... good, they won't live in london

[NKM]

*Something really stupid*
- Cenk Uygur

[calvinjackson]

He is such a great speaker. He delivers his thoughts in complete and well thought out sentences that just flow without the usual breaks with well huh and other filler sounds and pauses as he tries to think. He knows what he wants to say and he just says it! Coherently and clearly!! I am so impressed.

[gbone]

You were extremely spot on and ahead of your time on this, even if it was only 6 years ago

[Caveman_Conor]

Ted Talk: “can we build an AI without losing control over it?”
“No.”
*credits*