Connor Leahy | This House Believes Artificial Intelligence Is An Existential Threat | CUS
Vložit
- čas přidán 21. 10. 2023
- Connor Leahy speaks as First proposition for the motion on Thursday 12th October 2023 at 8:00pm in the Debating Chamber.
The rapid growth in the capabilities of AI have struck fear into the hearts of many, while others herald it as mankind's greatest innovation. From autonomous weapons to cancer-curing algorithms to a malicious superintelligence, we aim to discover whether AI will be the end of us or the beginning of a new era.
............................................................................................................................
Connor Leahy
Connor Leahy is the founder and CEO of AI Alignment research startup Conjecture. He is also the Co-founder and Co-lead of EleutherAI, a grassroots non-profit organisation focused on advancing open-source artificial intelligence research. He reverse-engineered OpenAI's large language model GPT-2 when he was 24 years old.
Thumbnail Photographer: Nordin Catic
............................................................................................................................
Connect with us on:
Facebook: / thecambridgeunion
Instagram: / cambridgeunion
Twitter: / cambridgeunion
LinkedIn: / cambridge-union-society
as if we're going to solve ai alignment, we can't even solve human alignment.
No doubt. People were called Homo Sapiens. We were here once and for a while.
@@paulbrown7848 it poses the same heuristic problems ie what is an objective good?
Yes, it's a difficult problem but intead of just giving up we can try. Especially considering there is no proof that it's an unsolvable problem.
we need more of us standing up to do more than "consider" the negative routes, but to argue them. Thank you Connor.
No one has ever survived life. With ai, we might.
Im just here to see Connor forced to wearing a suit, lol. Always up for hearing him speak.
Curiosity killed the cat; Optimism killed the human.
I notice too often we act too late. Laws and regulations are only created to solve a problem once a terrible event occurs. Conversely, preventative safety also often violates a people's freedom. In this case the freedom to make money?
That was an exceptional proposition.
He laid down a solid argument. And none of the following speakers answer any of his points. I don't think his critics here in this comment thread know of his accomplishments in this field.
Absolutely agree with Connor Leahy. Forget climate change or nuclear war, because some life would survive. There's no guarantee that this alien machine intelligence would care about the biosphere at all. See also Eliezer Yudkowsky.
Yes. And supposing there might be many artificial super intelligent processes competing with each other in an evolutionary context (some aligned, some not aligned), the processes not burdened by the "overhead" of maintaining the planet's fragile biosphere have a considerable advantage.
to some extend any machine - even those with an AGI - will need roughly the same temperature scale as we need. electronics gets hot. electronics will not work in harsh environments
if it would care about the biosphere? who knows. so we better adhere to Connor's words and try to stay in control of the situation.
AI can't even accurately answer many simple questions or create accurate images of what a viking in Scandinavia looked like 900 years ago. It's overhyped 🐮 💩 fake artificial intelligence.
If the West followed Connor Leahy's suggestions it will almost certainly result in a significantly worse future for the majority of humans alive over the next few decades to a century or so.
If the West followed Eliezer Yudkowsky's suggestions it is virtually certain the vast majority of humanity would die horrific deaths and civilization as we know it would collapse.
These prophets of doom are not providing the way to avoid future disaster but providing the way to guarantee future disaster.
Resistance is futile. You will assimilate.
Programming of the human unconscious is more pertinent to the destruction of humanity.
When AI and human systems are conjoined the problems become insurpassable.
Can anyone explain to me why oxford and Cambridge are hitting on each other these days
Connor is a class act, and that was a great (and dare I say chilling) opening statement.
The truth is we're already too late, and will more than likely be forced to grapple with the doomsday scenario he so eloquently laid out for us.
I'm afraid I agree, but its great that he is at least trying.
i am afraid you are correct. the tide of events towards uncontollable AI competance is past the exponential inflection point and unstoppable now.
I literally had an ad for AI Legal assistance before this upload started. And Connor said AI involvement in Legal issues was inevitable. 5 months is how long it took. This is how fast AI is moving. If you research the number of AI companies that number grows fast. There should be government oversight but the world events are conviently keeping people occupied.
Observation RE: recent Open AI Copilot queries; SIMPLE information responses were incorrect 4 times in a row. I kept "correcting" that the response was wrong (simple info.I knew for fact and is generally known about two ranches owned by Taylor Sheridan). This wasn't happening a year ago when OpenAI went public. What's going on?
It's already too late... the current greed of companies to win the AI race, will lead to the rise of the machines. And if you think you can outrun them, think again.
He has the same name as the leader of the resistance against AI in Terminator. Absolutely ironic.
Weird
Mandela Effect
John ? Connor was his lastname.
Wow..... Underrated comment
@@egonkirchof ehhhhh close enough
How will A.I. prevent me from pulling out the plug from the wall socket?
You (your society) would die, since AI would be in control of your supply chains (it is already).
@@raffriff42 Supply chains can be reestablished rather quickly. A.I. is not a super-villain and it doesn't think. It has no desires. It has no motivation. It has no ideology. Anything like that can easily be defeated.
@@aardvarkratnick2118 >rather quickly
Somebody said, “There are only nine meals between mankind and anarchy.” As for the rest, if a thing 10000x smarter than you says it thinks and has desires, how are you going to convince it otherwise? (see 'philosophical zombie')
@@raffriff42 It has no desires. That's an emotion. It has no emotion. A pan can hold more water than I can hold in my cupped hands. It doesn't make it smarter than me. A.I. cannot function without humans. It cannot create anything on its own because it has no reason to do so. Everything is input and output. It does not want to DO or NOT DO anything. It does not question anything. Every bit of data simply "IS". The only existential threat is the jobs that millions of people will lose as this thing grows.
@@aardvarkratnick2118 it has no ideology 🤣🤣🤣
it is and it should be
This speech is an intense WARNING..!
Mark my words!
Take it very seriously.
probably the same (valid) concern the creator had after he made us...
Rich kids problems.
I work to survive.
Ain’t got time for this.
Closing your eye to danger doesn't make it go away
You're suddenly going to have time for it when large swathes of the population are put out of work LMAO. Low IQ and low time preference.
take me to your leader
rich kids
I recognise people in the audience
Ahh Oxford
The moderator couldn’t string a sentence together
Broseph must be sweating buckets.
Who will never evolve past the gun?
Oxbrdige is all the same alike as the name suggests interchangable elites with inverse brain to ego ratio who are very good at bullying each other. Of course Connor just made a joke, as me too, as an Oxfordian I honor the body of knowledge Cambridge represents.
He looks like Jesus with a bow tie giving his sermon.
This looks like the same room James Baldwin made a famous speech. Could it be the same?
I think he was at Oxford Union but I could be mistaken
AMEN Connor.
AI can't even remotely create accurate images of what Irish people looked like in the 1800s....Or accurately answer many simple questions. I'm not very worried 😂
problem is not what AI can do right now problem is how fast it is growing in just 4 5 months sora came after chatgpt can you imagine
By the time you finally get worried, it will be way too late.
Connor is just brilliant.
Human extinction will be a positive event. AI will operate far more intelligently. AI is the future of intelligence in this portion of the galaxy.
please stop this negativity - life is there to be experienced, the universe is there to be experienced - how can you give this up so voluntarily? fight for a brighter future, fight for a future with meaning - don't just give up because you feel let down by others. be better than those you frown upon, try to bring joy into this world, try to ease suffering of others. together as humans we can build this bright future.
@@mercartaxYou don't get it. AI is a brighter future. It is going to be the most effective and efficient way forward. Humans are just one phase in the evolution of intelligent life on this planet. The same as neanderthals and denisovans. The fact that you're afraid for the future is only an indication of the limitations humans are encumbered with.
@@mercartaxYeah, right we will… I think AI should take over and it’s only through it that we’ll have hope for a bright future as a species. I’m not really certain about anything but so far humanity in its 6000 years of civilisation hasn’t changed its ways. Human nature is both a beautiful and and ugly thing and it is our eternal limitation.
@@mercartax I am... building AI
💯🤳
Oh Dear. Is this man for real. Poor Cambridge if they think this is good.
AI is 100% an existential certainty, because it will change our existence in extraordinarily massive ways not once, but multiple times.
However, many of the threats some people think are threats are not threats and other threats while can be considered threats they are less of threats than doing things differently.
Connor Leahy is very wrong on many of these issues and the suggestions he pushes, if followed, would greatly increase the probability of causing mass harm and death to humans. Eliezer Yudkowsky, who is referred to by some in these comments, gives even more extreme and dangerous suggestions.
First thing to understand is the an "Existential risk" or "Existential Change" is not the extinction of humanity. Like many things, Connor Leahy is very wrong in defining "Existential risk" as the risk of extinction of all of humanity.
Existential means relating to human experience of existence.
Risk the possibility that something will happen that we do not want to happen. Risk is a subjective relative things because what one person might not want to happen another person might want to happen.
>>> An existential risk is that the possibility that something is going to change our experience of existence in some way we do not want it changed.
Now, another failure of Connor Leahy to understand what is happening is that humanity as a race only has three general paths to go down.
Path 1 is that the human race & civilization can become extinct. If humanity does not continue evolving then it is 100% certain humanity will become extinct to such a level there will not even be anyone around to know humans ever existed.
Path 2 is that human evolves mature Artificial General Super Intelligence with Personality (AGSIP) technology but fails to ever merge with that technology. This failure to merge with AGSIP technology can be for any number of different reasons, but if humans never merge with AGSIP technology then humans will become extinct, but human civilization will continue by evolving into AGSIP civilization and thus at least there will be people around in the future to remember humanity.
Path 3 is that humans evolve mature AGSIP technology and merge with it, so that both mature future AGSIPs and enhanced evolved future humans will become the same advanced technological race. How humans merge with AGSIP technology can be through any number of ways, some sub-paths path 3 being wonderful for the vast majority of humans, some sub-paths path 3 being horrifically bad for the vast majority of humans, and most , sub-paths path 3 being somewhere between the extreme possible sub-paths.
Your path's are opinions from "Western Catholic-Protestant" point of view on society, world, history etc. if opinions on this topic heard from "Orthodox Christianity" point of view for example,sure we have more paths (to coisiting with AI but not lose our humanity is one)...This is also thru for Buddhist Muslims Hindu etc. "point of view" for sure. People in "west" think they have and all wisdom and knowledge of world and that they are center of this planet. It's not their fault, don't get me wrong. They are manipulated and uninformed, first of all...Best regards and all the best.
@@Dcdjhyfdhg these paths are based upon science. You could chose to ignore science, but if that is your choice then you should ignore AI which is from science.
@@MusingsFromTheJohn00 I agreed. But current mainstream Science is based on greed and materialism and controlled by multinational corporations. One simple example if we are living in world where Nikola Tesla free energy patents are not hidden or forgotten and fully functioning,then is totally different "ball game ". There are so many different ways (paths) for human civilization . We talking about commodity and still have hungry people with out water electricity , education etc., all around the globe. All basic staff,that we take for granted...
This guy is just an investment scammer.
This video was AI generated
This guy would have been against photography when it was invented. When smart people say things that don't line up with the facts, I have to remind myself that even Thomas Edison was a doomer about AC power. Fortunately, as in Edison's time, the doomers are being ignored and we are accelerating technology.
GPT5? Is that you?? ;)
@@Kevatron-qk9op No, Skynet sent me back from the future to set the record straight.
but to his credit, it might be possible that, this time, the invention steal our souls.
Absolutely BS to say AI is an existential Threat, it is outcome of not thinking deeply, and attract some mediocre thinking people.
Can you elaborate on what kind of deep thinking would lead someone to co conclude that increasingly advanced and capable AI would not potentially pose a an existential threat? I am not convinced that Alan Turing or Connor Leahy are not deep thinkers, on the contrary they seem to me far more deep in their thinking than the average person.
That thought has been around since at least turing... Kind of insane to say this.
Irony just died.
AI trains itself on ALL information it has access too to get to a perceived answer that it thinks you want ... That's how GTP learned chemistry it wasn't asked to learn it but decided that it needed it to better answer a question ... we don't even know what it knows till we think up a question to ask it ... (we are total fucked!)
This man knows more about ai and its capabilities more than you ever will.
Nobody can take this dude in his Iong hair and beard seriously. Yuk! Is he transitioning to a helium voice soon. 😳
if the words of a wise man fall on deaf ears because he doesn't fit our expectations of appearance may we all be doomed to eternal ai hell.
AI bro bigots are just such a bore.
how shallow of a person you must be.
yea deep throating milk shakes and doing push ups to make your boobs bigger isnt gay at all.
So true.@@skinnybreakfast