Every month, a Manifold poll is being posted to gain the consensus view of progress towards AGI. Unlike many of the markets and questions on Manifold, these polls will simply ask "has weak AGI been achieved," rather than providing a definition of the term. The goal is that given that the term is so broad, the only way to characterize AI progress is also to allow respondents to vote based upon their own understanding.
The questions only state that the poll does not require that the software system be able to manipulate the physical world. The polls open on the 23rd of each month and close on the 30th.
This market will resolve to YES if the January 2025 poll results in a ratio of YES/(NO + YES) greater than 0.57. The "no opinion" respondents will be ignored. Otherwise, this question will resolve to NO.
As a baseline, there has been a consistent linear trend from the May 2024 poll (18%) to the December 2024 poll when a plurality of respondents declared the achievement of weak AGI (50%).
RELATED:
1,000
3.00New related market: /SteveSokolowski/will-270-of-manifold-respondents-ag
Also, if this current market resolves YES, the clock will start ticking on the following market: /SteveSokolowski/will-six-months-or-fewer-elapse-bet
@stardust I know, that's weird, isn't it? I now use o1 pro as a doctor (except when I need to follow the letter of the law to get drugs prescribed), a lawyer, a financial advisor, a D&D campaign writer, and I have yet to find a case where humans perform better in any of those areas than the model does.
@SteveSokolowski Not your provided cases, but three areas where I find AI still to be weak are music (although a lot of modern music is degenerate too), philosophy, and debating theology, where obviously you wouldn't expect it to be a Jay Dyer, but yeah, ChatGPT argues theology like a woman (may God keep it that way).
For any STEM field that isn't active research I truly think us humans have been obsoleted. Honestly, I wonder what proportion of programmers use solely ChatGPT. I would guess the number to be higher than you think!
My take is that we've had "weak AGI" since GPT-4o. If o1 isn't "weak AGI" then I don't know what "strong AGI" is supposed to look like.
@stardust For music, I think the path forward is going to be humans that are able to abstract music creation to agents. General intelligence models unassisted are quite poor at writing music.
Right now, I either have to input notes and chords on a keyboard, which is time consuming, or use models, which often outputs stuff that I can see obviously how to improve but can't. That can be fixed by using traditional music software, but telling the agent "transpose this from F# minor to C# Dorian mode" or "add a distortion guitar countermelody with the following characteristics." This is how you get superhuman music; contrary to what picky Manifold commentators say, I believe the AI music I can create already far surpasses the average band. The only thing limiting this right now is the expense of setting up an agent, and this is the workflow I envision getting the first song on the radio, not directly finding the best music model prompts.
I don't agree with you on poor philosophical arguing, if that's what you mean. The models seem to actually have quite a good understanding of how they operate and of human emotions and experience. You can have interesting conversations with o1 pro about philosophy, although perhaps if I did that for tens of hours I would find otherwise.
One issue the models are also poor at is serving as a therapist, which is a shame, because psychologists are probably the most in-demand specialty right now. My mom needed one and simply was never able to get one because none of them returned her calls, even when I said I would beat their normal prices.
@SteveSokolowski Well, you can share your music and I'll take a listen to it. For what it's worth, I don't think comparing to the average band is of much worth. I would definitely bet that AI could create less degenerate music than modern musicians, but I would bet against AI for the foreseeable future being able to outperform musicians from 500-1000 years ago.
The models generally repeat the secularist consensus with a whole bunch of words, whole bunch of nothing. Generally it's very easy to "convince" the model into "your position". It can serve as an encyclopedia I guess, but will never back up the ideas it cites very strongly at all.
It's regretful that such a thing happened to your mother and I will pray for her. I will not push it, but know that the Orthodox Church is always, at any time, free of charge, open to anyone with a spiritual problem facing their life. May the LORD have mercy
@stardust https://soundcloud.com/steve-sokolowski-797437843/six-weeks-from-agi
Scientists almost always aren't willing to consider other philosphical theories, like any that are founded on the premise of "consciousness first." This is despite the obvious flaw that any theory which does not explain what occurs to consciousness "before birth" or "after death" (if time even exists) is fundamentally flawed.
Those two questions need to be answered in the next theory, because the Standard Model has gone as far as it can go. It works well for producing a mathematical model of the universe, but cannot unify itself with quantum mechanics and consciousness. My prediction is that we will indeed be able to answer those questions within the next few years, once superintelligences get a better understanding of what reality is.
@SteveSokolowski I don't really like the song, in fact, I think it's quite bad. That said, I do think you've accomplished your goal of being around as good as the average modern professional musician with AI.
As Christians we have a very clear answer: the existence of the immortal soul, which has a real ontology and doesn't collapse into mereological nihilism like physicalist theories.
I don't know if you saw but awhile ago when I was debating a liberal here I had my critiques of the Standard Model too. According to the standard model all objects in the universe that are physical possess mass. All objects in the universe are compounds of atoms, which are compounds of protons, neutrons, and electrons. All protons and neutrons are made of quarks and gluons and all electrons are just excitations in the electron field.
Mass arises when
1) Quarks and gluons interact
2) Quarks interact with the Higgs field
3) Excitations in the electron field interact with the Higgs field
But quarks and gluons don't exist outside of protons and neutrons. Quarks and gluons are the protons and neutrons respectively. Mass arises when 1, 2, and 3 are met and prior to mass the physical universe cannot be said to exist. This interaction always occurs on the fundamental level, so it must then be said that quarks and gluons do not possess mass on their own, likewise excitations in the electron field do not possess mass on their own, and neither does the Higgs field possess mass on its own. For mass to arise, the interactions from 1, 2, and 3 must always occur, otherwise the physical universe does not exist.
But this interaction, which binds these particles together, does not exist physically. Rather, the arisal of mass is contingent on this interaction having occured, otherwise there are no particles to speak of -- that is, protons and neutrons and electrions respectively -- as without these interactions these objects would not give the mass necessary in an atom to exist within spacetime. This interaction, therefore, must be understood as a law of the universe. It always occurs, if it doesn't, nothing else occurs.
Empirical evidence that which the scientific method relies on depends on measurement and so forth. Non-physical things like this law cannot be measured, and because they are not made of a substance (or the law would be contingent on itself) they cannot be observed through the senses, because the senses do not perceive things that have no being in the universe. Therefore, this is a dead end for empirical inquiry. The physicist must necessarily admit that empirical inquiry ends at the point of generation of mass, as anything beyond this event horizon is non-physical and non-empirical in nature. So, the only explanation left for an atheist is that the universe does not exist, because its substance (that which science makes inquiry into) is contingent, according to science, on something that is beyond science. This law.
This law is thus, from the atheist's perspective, nomological because it cannot be explained scientifically. It can only be asserted as existing and cannot be proven, in other words. This is a direct contradiction to logic, which contains the law of the excluded middle and law of non-contradiction. The proposition "the universe exists" cannot be answered by a nomological law because that's the middle that's excluded. The universe either exists or it doesn't. In order to validate the evidence that has taken you to this dead end, you must assent that the universe does exist, otherwise your "research" means nothing, so you must also assent that this law is true, whereby that calls for evidence in respose to the proposition that "the universe exists".
The Orthodox Christian God, Father, Son, and Holy Spirit, is transcendental. He is timeless, atemporal, eternal, non-physical, non-dimensional, spaceless, as is His act of creation. From direct logical necessity, God exists to account for the proposition "the universe exists", as we have demonstrated above.
When the atheist asserts this "nomological law", he is essentially asserting God.
@stardust Consider what you said there - in a few months of learning and research, and at just $30 for this song, I was able to outperform 50% of professional musicians without spending $20,000+ to hire a band and vocalist and do it the old way.
If interaction between particles is the base law of the universe, then that in itself implies that consciousness is the basic reality, doesn't it? It's not necessary to go any further than that. Interaction is an active process, whereas particles existing without interaction is a passive one that could occur without anything conscious existing.
While I would not necessarily extend this all the way to traditional Christian beliefs, I do think that Catholicism and Orthodox Christianity have gotten some of the way there. My guess is that at some point, when scientists began discovering mathematical laws that explained the physical universe, they, as all humans do, thought that their theories were better than everyone else's. That crowded out the prevailing religious view.
What should have happened is that the religious view would be corrected, by replacing the parts that science showed to be incorrect while retaining ideas that had guided humans for a long time before.
The big problem with religion today is that it is not considered a scientific subject like other things. People treat the concept of God as something that is "unprovable" and human birth and death as "unknowable." There's no reason at all to suggest that the concept of God is not something that can be researched and adjusted as scientific evidence develops.
It's nonsensical to suggest that there are two "realms" - a physical realm and a spiritual realm. There is one reality, and as humans learn more about it, this distinction should disappear.
@SteveSokolowski Well yeah, but I don't know if that's so much a statement on AI's music-writing capabilities in this case as it is on modern music's degeneracy. If we were comparing to the music of 500 years ago, or even 50 years ago, it would be a very different story. Right now where celebrities just churn out slop it's a perfect environment for an AI to thrive. I actually wouldn't even be surprised if a few of them are using AI already.
If interaction between particles is the base law of the universe, then that in itself implies that consciousness is the basic reality, doesn't it? It's not necessary to go any further than that.
Well no. Even if you shoehorn consciousness into "interaction between particles", which I believe is ill-founded and ill-advised (mereological nihilism follows), my point is that this is a dead end for empirical inquiry. Empirical inquiry can only study the physical. The physical is that which has mass. Mass is generated through a set of laws. These laws are not physical. The physicalist atheist, who must preserve that "the universe exists", must therefore self-refute by asserting the existence of a nonphysical nomological law which guarantees the ontology of the universe.
In this case all you'd be doing is admitting from step 1 (if you weren't to admit God) that consciousness actually has no basis other than "it must exist, because it would be absurd if it didn't", which is true, but not a justification.
As Christians we admit the existence of the soul.
While I would not necessarily extend this all the way to traditional Christian beliefs, I do think that Catholicism and Orthodox Christianity have gotten some of the way there. My guess is that at some point, when scientists began discovering mathematical laws that explained the physical universe, they, as all humans do, thought that their theories were better than everyone else's. That crowded out the prevailing religious view.
This is the fault of the Roman Catholics. The truth of the matter and that which modern man ceases to recognize is: what is science based on? How can you know science works or is good? What are the preconditions for "science"?
The Christian answer has always been God. If you get a "scientific answer" that contradicts God, one of three things are true
1) We misinterpreted God somewhere, this is possible for things which are not doctrine and just merely common belief.
2) The "scientific answer" as stated is wrong, as doctrine prevails. Similarly, "science" could never disprove that 2+2 is 4. Note that doesn't have to mean that the facts that are observed are false, just that the overall narrative is not correct.
3) Someone is lying about something.
In any case, the idea that science can revise God's word is a deeply Protestant and Roman Catholic notion; much like it cannot revise math, this is impossible. The very easy thing to do from the religious side is just... not to say things which are false, which is naturally hard for RCs and Prots as they are heretical.
The big problem with religion today is that it is not considered a scientific subject like other things. People treat the concept of God as something that is "unprovable" and human birth and death as "unknowable." There's no reason at all to suggest that the concept of God is not something that can be researched and adjusted as scientific evidence develops.
This is another Protestant problem. The idea that "God is unprovable" is ridiculous given that His existence is necessary for logic, ontology, knowledge, etc. and as Orthodox Christians that is what we have held to. The idea that God is completely unknowable is also directly heretical; His essence is, His divine energies are not, and to insist that they are would be to invoke the Neoplatonic monad, to essentially state that God is not a personal God, and to devolve into absurdity.
We may know and gain information about He, the LORD, through the Holy Orthodox Church.
It's nonsensical to suggest that there are two "realms" - a physical realm and a spiritual realm. There is one reality, and as humans learn more about it, this distinction should disappear.
You are correct that there is only one reality. That said, that does not imply the former. Absolutely non-physical things must be admitted, otherwise what justifies the physical? It can't "just be".
There have been three previous polls similar to this, and all of them have resolved against the consensus of the market - one of them was well below 20% at one point.
Manifold, like everyone else, continually underestimates technological progress.
