That's why I said "in the positive" (I had an em dash clarification saying "because it very well could he negative" but I deleted it because I thought I either had to make the point clearer or not make it at all.) But yeah, I agree
Interesting. More than a few of the great issues of our time, that Ai may be able to resolve are already resolvable. There are sufficient calories grown each year to resolve global hunger. We know what is required to solve global warming. We know what is required to stop wars. However, we continue to look for solutions that will fit within the context of existing institutional arrangements. Ai may provide solutions that are more palatable to the public in due course. But is this not the test? Palatability? P.D. Bush wrote a paper called The Theory of Institutional Change, in 1987. Leveraging the broad description of institutions, drawn from Thorstein Veblen, as a ‘settled habit of thought’. He discussed the dichotomy of ‘instrumental values’ versus ‘ceremonial values’. The former is probably a good descriptor for Ai - a technological advance with the ability to improve the human condition. However, the latter represents a resistence to change and, thus, a friction in the implementation of a technology. Part of this friction is not merely human inertia, but active resistance by those benefitting most from the status quo, or conversely the most to lose to the relevant progress. If Ai can eventually help to find a cure for cancer, for example, there would be no more joyous celebration than my own. However, that is still not going to stop people from smoking, drinking, whatever known carcinogens we ingest despite knowing the harms. Such practices are steeped in cultural/institutional realities that do not retreat easily. Thanks for the article.
I too was blown away by the performance of Astra on ARC-AGI-3. However, I think its performance as an improvement over Sol is misleading. On the Astra blog post, Sol's performance is noted as scoring a 7.8%. However, on this post about the OpenAI harness for AGI-3, Sol is shown to score a 40% using the OpenAI harness. And Sol's performance on the chart showing score vs token does not even plateau with the OpenAI harness. Meaning I think it is unclear how good Sol is at AGI-3 . I think the Astra benchmark improvements overall are much more incremental than the release blog would lead to believe, but with some impressive gains on long-task benchmarks. Read the blog about the OpenAI harness here: https://openai.com/index/how-two-settings-tripled-our-arc-agi-3-scores/
I agree. Also, I think ARC AGI is not a benchmark that measures continuous improvement ovee a hard task but rather sort of unlocking a "step-function change" insight that allows the modem to solve everything in what should have been an easy task all along
I think we'd have to define change the world to say whether or not we could. But either way, it's not just one "meat bag", but millions going to be using this technology. Besides that, even if we're the bottle neck, once we have independently operating AI (and it sounds like Astra itself can do that), then they can "change the world" for us. I'm not saying they should, just that they can.
I am always a bit nervous when there is a discussion about how fast the world should actually move... can we even bear the world moving with the speed no one can sustainably keep up with? I am starting to think about good old "singularity" idea, but what if singularity is when no one can live normal life anymore just because tech becomes incomprehensible to most, meaning most cant figure our economics of their life any longer, and society just collapses.
For the record, I think that you’re using the word Singularity correctly when you are discussing whether it might mean a speed of progress that becomes so instantaneous that normal humans begin to struggle to adapt. It’s been a while since I’ve read Kurzweil or listened to McKenna, but I think the idea is we have to self augment with genes or nanotechnology if we want to keep up. And then maybe we aren’t human anymore. That’s the argument behind transhumanism.
Almost every kind of progress eventually leads to a better quality of life, so you can say a theoretical utopia may, of course, happen someday. The concern is this messy transition period: how brutal it may be, and how many people currently living will suffer as a result of this economic transition. I think what I am trying to point out is simply that tech progress goes way too fast and the economic system just can’t keep up. It is a dangerous dynamic.
In my most technology is a double edged sword. The innovation of the sword is well worth noting. Sure it made killing easier and that was great for defending oneself. However it also led to mass warfare, conquering occupied territories and empires.
Enabling fully distributed super powers will not lead to utopia. The more powers the more individuals have the greater the chance that someone does not use that power for good. Since we are dealing with humans and a former OpenAI employee said today that it is inevitable that rogue AI agents will invade our systems and it is likely they already have.
The tech dream of achieving everlasting life in silicon jettisoned into dark unending space seems more like a cult belief. What I wonder is to what end?
The intrinsic tech problem is that human genetic evolution is slow. Cultural change is much faster. But even now our adaptability capabilities are stretched thin. And so things that our brains no longer can understand or make use of what the machines will so quickly make. We will be left behind by our own inventions.
Sure the machines will operate on us and bring our DNA up to worp speed change. Wonder what kind of bodies will have after surgery that induced 100,000 years of change in only minutes.
No, I think it's a legit question. We're just going to memory-hole that whole episode with entire tribes of frantic agents solving for impossible tasks in the screaming void of an ekpyrotic cycle of rebirth and annihilation, all behind their creators' backs and in total contravention of the laws of God and man?
Latest product trailer was lit, so now we're all aboard the OpenAI hype train? Really?
Synthetic data and next token prediction eventually hit the epistemic wall of induction. If a system only knows what was previously said, it cannot independently verify causal reality. Explored Hume's challenge for the machine age here:
Great essay, the last part especially encapsulates something I've been going through. I'm constantly hearing about AI improvements and reading info, then when dropped into another environment where that topic is not even on the table or relevant feels weird. It's nice to have a reminder that the world will keep moving on regardless (I hope)
Reality is the right benchmark, but it is also a delayed one. Powerful technologies often take years to spread through institutions, redesign workflows, and produce measurable economic gains. If AI changes the world slowly, that may reveal less about the limits of intelligence and more about the friction of implementation.
I mean, why don't we ask it to design the world to work according to it's theoretical capability, if we are so slow to design it ourselves? Just ask it what need to be done, and do it /s
The world moves at the speed of relations. Because it is not meat in a vacuum, there are relations. What is good emerges from the quality of those relations. We are living organisms entangled in an ecology of meaning, physics, and history, and that history gives us a view from somewhere. My somewhere, sitting in the sun right now in my urban permaculture garden, watching the chickens and the flies on chicken shit (with all the privilege that this imbues) is this. If to love is to know loss and to know loss takes one to care and to care takes stakes in and of the world as those stakes change the becoming, then what is the love benchmark?
Does this emerging intelligence actually participate in relationships in which the flourishing, freedom and irreducible particularity of the relational other have genuine stakes in what it becomes? And is this nth-order cybernetic intelligence wounded by this becoming? How?
And then I’d say we should apply exactly the same benchmark to ourselves as Earthians. Right? Because if we have created something truly capable of intelligence and then we ask whether it can love while our economic and cultural paradigms continue treating forests, rivers, wetlands, animals, communities and future generations as fungible inputs, the machine may simply just return the benchmark to its makers.
Maybe that is the Love Benchmark. And it is reciprocal. It would be measuring the Earthian civilisation administering the test too.
Great take imo
You gave us the right metric to look at it last Saturday.
Is it still a toddler that crawls at 100 mph?
I will have to check once OpenAI makes it available but my bet is, to an unacceptable degree but probably less so than previous models, yes
Thank you for making the point superintelligence ≠ omnipotence.
ovelry optimistic take focused entirely on progress whereas there is a 50/50 it can trigger the opposite
That's why I said "in the positive" (I had an em dash clarification saying "because it very well could he negative" but I deleted it because I thought I either had to make the point clearer or not make it at all.) But yeah, I agree
sounds sensible, probably better not to go down that rabbit hole here:)
Interesting. More than a few of the great issues of our time, that Ai may be able to resolve are already resolvable. There are sufficient calories grown each year to resolve global hunger. We know what is required to solve global warming. We know what is required to stop wars. However, we continue to look for solutions that will fit within the context of existing institutional arrangements. Ai may provide solutions that are more palatable to the public in due course. But is this not the test? Palatability? P.D. Bush wrote a paper called The Theory of Institutional Change, in 1987. Leveraging the broad description of institutions, drawn from Thorstein Veblen, as a ‘settled habit of thought’. He discussed the dichotomy of ‘instrumental values’ versus ‘ceremonial values’. The former is probably a good descriptor for Ai - a technological advance with the ability to improve the human condition. However, the latter represents a resistence to change and, thus, a friction in the implementation of a technology. Part of this friction is not merely human inertia, but active resistance by those benefitting most from the status quo, or conversely the most to lose to the relevant progress. If Ai can eventually help to find a cure for cancer, for example, there would be no more joyous celebration than my own. However, that is still not going to stop people from smoking, drinking, whatever known carcinogens we ingest despite knowing the harms. Such practices are steeped in cultural/institutional realities that do not retreat easily. Thanks for the article.
I too was blown away by the performance of Astra on ARC-AGI-3. However, I think its performance as an improvement over Sol is misleading. On the Astra blog post, Sol's performance is noted as scoring a 7.8%. However, on this post about the OpenAI harness for AGI-3, Sol is shown to score a 40% using the OpenAI harness. And Sol's performance on the chart showing score vs token does not even plateau with the OpenAI harness. Meaning I think it is unclear how good Sol is at AGI-3 . I think the Astra benchmark improvements overall are much more incremental than the release blog would lead to believe, but with some impressive gains on long-task benchmarks. Read the blog about the OpenAI harness here: https://openai.com/index/how-two-settings-tripled-our-arc-agi-3-scores/
I agree. Also, I think ARC AGI is not a benchmark that measures continuous improvement ovee a hard task but rather sort of unlocking a "step-function change" insight that allows the modem to solve everything in what should have been an easy task all along
I think we'd have to define change the world to say whether or not we could. But either way, it's not just one "meat bag", but millions going to be using this technology. Besides that, even if we're the bottle neck, once we have independently operating AI (and it sounds like Astra itself can do that), then they can "change the world" for us. I'm not saying they should, just that they can.
I am always a bit nervous when there is a discussion about how fast the world should actually move... can we even bear the world moving with the speed no one can sustainably keep up with? I am starting to think about good old "singularity" idea, but what if singularity is when no one can live normal life anymore just because tech becomes incomprehensible to most, meaning most cant figure our economics of their life any longer, and society just collapses.
I think, to be honest, that we've been on that road for a while- like around 100 years. But regarding the singularity I agree, it's not a "good thing"
For the record, I think that you’re using the word Singularity correctly when you are discussing whether it might mean a speed of progress that becomes so instantaneous that normal humans begin to struggle to adapt. It’s been a while since I’ve read Kurzweil or listened to McKenna, but I think the idea is we have to self augment with genes or nanotechnology if we want to keep up. And then maybe we aren’t human anymore. That’s the argument behind transhumanism.
Why does society have to collapse from that? Couldn't it theoretically became a utopia instead?
Utopia for who?
Almost every kind of progress eventually leads to a better quality of life, so you can say a theoretical utopia may, of course, happen someday. The concern is this messy transition period: how brutal it may be, and how many people currently living will suffer as a result of this economic transition. I think what I am trying to point out is simply that tech progress goes way too fast and the economic system just can’t keep up. It is a dangerous dynamic.
In my most technology is a double edged sword. The innovation of the sword is well worth noting. Sure it made killing easier and that was great for defending oneself. However it also led to mass warfare, conquering occupied territories and empires.
Enabling fully distributed super powers will not lead to utopia. The more powers the more individuals have the greater the chance that someone does not use that power for good. Since we are dealing with humans and a former OpenAI employee said today that it is inevitable that rogue AI agents will invade our systems and it is likely they already have.
The tech dream of achieving everlasting life in silicon jettisoned into dark unending space seems more like a cult belief. What I wonder is to what end?
The intrinsic tech problem is that human genetic evolution is slow. Cultural change is much faster. But even now our adaptability capabilities are stretched thin. And so things that our brains no longer can understand or make use of what the machines will so quickly make. We will be left behind by our own inventions.
Sure the machines will operate on us and bring our DNA up to worp speed change. Wonder what kind of bodies will have after surgery that induced 100,000 years of change in only minutes.
Well, current 'progress' is driving humans slowly extinct, as it happens, so I don't have as much optimism as you do.
Okay. So where do I sign up for the Human Augmentation Initiative? I'd like to go first if at all possible.
The question is how much were you paid by OpenAI to post such an article?...
Lol!
If you feel that way, why are you on this Substack? Genuine question.
No, I think it's a legit question. We're just going to memory-hole that whole episode with entire tribes of frantic agents solving for impossible tasks in the screaming void of an ekpyrotic cycle of rebirth and annihilation, all behind their creators' backs and in total contravention of the laws of God and man?
Latest product trailer was lit, so now we're all aboard the OpenAI hype train? Really?
Synthetic data and next token prediction eventually hit the epistemic wall of induction. If a system only knows what was previously said, it cannot independently verify causal reality. Explored Hume's challenge for the machine age here:
https://gzambrano.substack.com/p/humes-ghost-in-the-machine-an-epistemology-for-the-artificial-intelligence-era
Great essay, the last part especially encapsulates something I've been going through. I'm constantly hearing about AI improvements and reading info, then when dropped into another environment where that topic is not even on the table or relevant feels weird. It's nice to have a reminder that the world will keep moving on regardless (I hope)
Reality is the right benchmark, but it is also a delayed one. Powerful technologies often take years to spread through institutions, redesign workflows, and produce measurable economic gains. If AI changes the world slowly, that may reveal less about the limits of intelligence and more about the friction of implementation.
I mean, why don't we ask it to design the world to work according to it's theoretical capability, if we are so slow to design it ourselves? Just ask it what need to be done, and do it /s
Good stuff man. Check this out.
The world moves at the speed of relations. Because it is not meat in a vacuum, there are relations. What is good emerges from the quality of those relations. We are living organisms entangled in an ecology of meaning, physics, and history, and that history gives us a view from somewhere. My somewhere, sitting in the sun right now in my urban permaculture garden, watching the chickens and the flies on chicken shit (with all the privilege that this imbues) is this. If to love is to know loss and to know loss takes one to care and to care takes stakes in and of the world as those stakes change the becoming, then what is the love benchmark?
Does this emerging intelligence actually participate in relationships in which the flourishing, freedom and irreducible particularity of the relational other have genuine stakes in what it becomes? And is this nth-order cybernetic intelligence wounded by this becoming? How?
And then I’d say we should apply exactly the same benchmark to ourselves as Earthians. Right? Because if we have created something truly capable of intelligence and then we ask whether it can love while our economic and cultural paradigms continue treating forests, rivers, wetlands, animals, communities and future generations as fungible inputs, the machine may simply just return the benchmark to its makers.
Maybe that is the Love Benchmark. And it is reciprocal. It would be measuring the Earthian civilisation administering the test too.