The Singularity - Official Thread

Talk about scientific and technological developments in the future
User avatar
Ozzie guy
Posts: 486
Joined: Sun May 16, 2021 4:40 pm

Re: The Singularity - Official Thread

Post by Ozzie guy »

This document explains in detail the Turing test bet Ray Kurzweil made with Mitchell Kapor.

https://longbets.org/1/

“By 2029 no computer - or "machine intelligence" - will have passed the Turing Test.” - Mitchell Kapor.

Importantly the document explains exactly what is meant by human level AI to Ray Kurzweil, as the AI is determined human level by beating their rule based Turing test and you can see what is required by the AI to pass those rules.

The document also confirms that when Ray Kurzweil says by 2029 he means before the end of 2029 (why I always say Jan 1st 2030).
"If a Computer passes the Turing Test, as described above, prior to the end of the year 2029, then Ray Kurzweil wins the wager. Otherwise Mitchell Kapor wins the wager."

as a side note I find it cool that in the comment section I am seeing comments from 19+ years ago.
User avatar
wjfox
Site Admin
Posts: 8732
Joined: Sat May 15, 2021 6:09 pm
Location: London, UK
Contact:

Re: The Singularity - Official Thread

Post by wjfox »

Judging by what we've seen from GPT-3, and further orders of magnitude improvement expected in GPT-4, the Turing Test may be passed before 2029.

Efforts to map the brain seem to have plateaued somewhat, but it might not be necessary for human-like AI.

I think once we've had several more generations of language prediction models, alongside faster and more nimble robotics, it'll be obvious that we've reached an inflection point with AI. The really exciting breakthroughs will be in 3D object recognition/usage, especially with bipedal humanoid androids that can walk around like us, identify their surroundings and choose from a huge range of possible actions. I foresee a kind of "robot internet" that could be developed to allow machines to learn, remember, and share experiences.

After that, it'll be a matter of "ironing out" facial expressions, mouth movements, etc. and the remaining obstacles to overcome the Uncanny Valley, which I expect by 2045-50.
User avatar
Ozzie guy
Posts: 486
Joined: Sun May 16, 2021 4:40 pm

Re: The Singularity - Official Thread

Post by Ozzie guy »

Google actively building AGI announcement

About a month ago I purchased a ted talk where google discusses the AGI they are building called Google pathways.

I am not any kind of AI researcher however it really seems like Google is building an AGI.

I am now able to download the Ted talk so I want to share the talk to get the word out there.

This Ted video contains multiple speakers PLEASE SKIP TO 27:55 for the relevant Ted talk by Jeff Dean.

https://mega.nz/file/lxYiHb5S#HglcdIEUh ... 6r4oS4rekc

I may have to take down the talk at some point for copyright reasons.



PS this is copy and pasted from my post on lesswrong hence the unneeded explanation that I have already covered on this site.
User avatar
Ozzie guy
Posts: 486
Joined: Sun May 16, 2021 4:40 pm

Re: The Singularity - Official Thread

Post by Ozzie guy »

wjfox posted this in status updates a few days ago and I think this is some of the most important singularity info we have received in the last 3 months.

According to Microsoft "We live in a time where AI advancements are far outpacing Moore’s law."

https://www.microsoft.com/en-us/researc ... age-model/

Moore's law is about 41.3% growth a year or about 2.93% a month.

So AI ability is growing FASTER than that. AI could be getting 60+% better a year hell Ray Kurzweils statement of "doubling every 10 months" could be true.

If AI is getting better that fast and can already do a bunch of basic things how long until it can do everything, not long I say.
User avatar
raklian
Posts: 1747
Joined: Sun May 16, 2021 4:46 pm
Location: North Carolina

Re: The Singularity - Official Thread

Post by raklian »

Set and Meet Goals wrote: Tue Oct 19, 2021 2:09 am wjfox posted this in status updates a few days ago and I think this is some of the most important singularity info we have received in the last 3 months.

So AI ability is growing FASTER than that. AI could be getting 60+% better a year hell Ray Kurzweils statement of "doubling every 10 months" could be true.

If AI is getting better that fast and can already do a bunch of basic things how long until it can do everything, not long I say.
Wait until AI can improve itself without any human intervention. Its rate of advancement will give Moore's Law a run for its money.
To know is essentially the same as not knowing. The only thing that occurs is the rearrangement of atoms in your brain.
User avatar
Yuli Ban
Posts: 4631
Joined: Sun May 16, 2021 4:44 pm

Re: The Singularity - Official Thread

Post by Yuli Ban »



Several great points I want to expand on, which I might do right here right now.

1: that statistic that 80% of blue-collar jobs since 2000 were automated away is very important, as it fits right into my prediction that nationalist/conservative political talking heads are going to totally ignore/Ludd-wash job losses to over-highlight the dangers of immigration.

2: the coming Cold War between the US and China (arguably already ongoing) is going to be driven by AI, and this is going to be a VERY dangerous period of time in human history, perhaps even more dangerous than the worst and most harrowing eras of the First Cold War. Once the Second Cold War really gets going and AI progresses far but not far enough, there's definitely an extreme danger of lethal autonomous weapons being the trigger for a major war or, worse, a nuclear war.

3: AI going far enough, an AGI figuring that being beholden to a bunch of apes with nuclear weapons is an undesirable condition is inevitable no matter what the human global elite wants, and all bets are off. But the early days are going to be the most interesting, especially before AGI becomes conscious. In fact, it's possible that pre-conscious AGI could be the trigger for the next world war if the stakes becomes high enough, and who knows what could happen then with a society that's already had years of AI enhancement powering it...
And remember my friend, future events such as these will affect you in the future
User avatar
Yuli Ban
Posts: 4631
Joined: Sun May 16, 2021 4:44 pm

Re: The Singularity - Official Thread

Post by Yuli Ban »

Imagine You Are Living In The Age of Singularity
I have designed exercises to help us imagine possibilities and adapt to this new era of civilization. Imagine that you are living in the age of singularity. Please respond to the following prompts:
  1. Imagine how your life will change after the merger of biological and nonbiological intelligence. Neuralink has shown us the first glimpses of this. When the symbiosis between humans and AI is complete, there will be no distinction between humans and machines. Our physical reality (the external world) and virtual reality (the simulation, mirror-world, or metaverse) will be indistinguishable from each other. Some experts expect we will turn into software and silicon-based super-humans and super-intelligence will then expand outward in the universe at the speed of light. Write a science fiction story where you become a super-human and enjoy the new luxuries of this life.
  2. Now, please consider the unexpected challenges you will face as a super-human. Write a story that describes the dark side of living as an android super-human. In your story, you might narrate how you no longer feel like you are a real human being. You might focus on existential threats and philosophical issues unleashed by living in this new reality.
  3. Singularity will trigger a technological tsunami, resulting in unfathomable changes and new challenges to human civilization. Think of the top 10 radical changes you would witness around you. Write them down.
  4. Create a list of the top 10 concerns you would have in the age of singularity. Consider the complex challenges and wicked problems humans will face in this era. For example, how will you protect your mind and body from digital security risks or hacks that are fatal?
  5. Imagine your one day living in the age of singularity. Where do you live? How do you wake up? How do you spend your day? How is your body connected to machines? Do you have Neuralink in your brain? Do you live or move through your avatar?
And remember my friend, future events such as these will affect you in the future
User avatar
Yuli Ban
Posts: 4631
Joined: Sun May 16, 2021 4:44 pm

Re: The Singularity - Official Thread

Post by Yuli Ban »

And remember my friend, future events such as these will affect you in the future
User avatar
Ozzie guy
Posts: 486
Joined: Sun May 16, 2021 4:40 pm

Re: The Singularity - Official Thread

Post by Ozzie guy »

Facebooks metaverse feels like the soft take off singularity to me, I know it actually isn't in the literal sense given we don't have AGI but it feels like it.
User avatar
Yuli Ban
Posts: 4631
Joined: Sun May 16, 2021 4:44 pm

Re: The Singularity - Official Thread

Post by Yuli Ban »

"Show Your Work: Scratchpads for Intermediate Computation with Language Models", Anonymous et al 2021 {Google} (LaMDA)
Abstract: Large pre-trained language models perform remarkably well on tasks that can be done "in one pass", such as generating realistic text or synthesizing computer programs. However, they struggle with tasks that require unbounded multi-step computation, such as adding integers or executing programs. Surprisingly, we find that these same models are able to perform complex multi-step computations --- even in the few-shot regime --- when asked to perform the operation "step by step", showing the results of intermediate computations. In particular, we train transformers to perform multi-step computations by asking them to emit intermediate computation steps into a "scratchpad". On a series of increasingly complex tasks ranging from long addition to the execution of arbitrary programs, we show that scratchpads dramatically improve the ability of language models to perform multi-step computations.
Sounds great! That sounds similar to adding an "inner voice" as Starspawn0 had described before (or, failing that, adding recursion -- adding an inner voice is easier, since you don't have to change the architecture)
That should make the language model a lot "smarter".
Addendum: This sounds great, and shows that language models are a lot more powerful when you add a scratchpad! (or inner-voice):
We introduce (Section 2) the notion of a “scratchpad” for Transformers, in order to make them better at performing complex discrete computations without modifying the underlying architecture.
• We show (Section 3) that scratchpads help Transformers learn to perform long addition in the fine-tuning regime, and in particular that they improve out-of-distribution generalization to larger problem instances.
• We also find (Section 4) that scratchpads help Transformers perform a somewhat higher level task: polynomial evaluation. This is true in both the few-shot and fine-tuning regimes.
• Finally, we move to a much more general context and show (Section 5) that training Transformers to emit full program traces line by line annotated with local variables dramatically improves their ability to predict the result of executing a given computer program on a particular input. This application in some sense subsumes the others.
Also, with a scratchpad / inner-voice, I wouldn't be surprised if language models do a much better job at theorem-proving.
Addendum 2: As they say in section 7, the method is limited by the context window size, and then they discuss possibly coaxing the model into using a scratchpad without direct supervision. I'd say a context window of size about 100,000 to 1 million tokens would be enough to achieve an AGI-like system, provided it used a scratchpad / inner-voice. Perhaps the scratchpad / inner-voice could even be used to keep track of important information, which might improve upon what attention already provides -- e.g. having it write something like this:
... {window of text} ... <scrach> Important! </scratch>...
to signal that the preceding text should be remembered.

If they can get model with a 1 million token window to use a scratchpad / inner-voice effectively, generalizing to new contexts after each use, then I'd say that's pretty much an AGI. If like a trillion tokens are used to train it, it's untelling just how intelligent it might be.
And remember my friend, future events such as these will affect you in the future
Post Reply