This is a silly thing to include on the list since it doesn't seem to be true?
Is there something I missed that shows it's all shenanigans?
It could be a Chinese drone with rat-cell-on-a-chip inside it
The guy's testimony was that somebody told him stuff and it seemed scary. Congress made a dog and pony show that a significant chunk of people attended.
Parse the details, not the overarching fantasy.
The government admitted to doing this in the 70s-90s to keep not-so-strange technology hidden behind The Strange. Trim with Occam.
The ticTac videos aren't interesting if you look at the limitations of perspective and remember that Air Force pilots aren't supreme authorities for objectively assessing parallax and assigning airspeed to all moving objects while flying a fucking cracker attached to a rocket.
They don't need to be "supreme authorities" on anything -- they have insane technology acting as their eyes and as their objective measuring and tracking systems.
2) The burden of proof is on those who make an extraordinary claim. There is a perfectly mundane explanation of the cause of apparently fast-moving object which is supported by available evidence. Even if that depends on some assumptions, you would have to demonstrate those assumptions to be false before a less likely alternative, i.e., that it was an extraterrestrial spacecraft, should be adopted.
To make the extraordinary claim more plausible, you have to rule out the more mundane possibilities. Prove it's doing something unusual. Until then, "we can't prove it either way" doesn't mean there's a 50/50 chance.
It's like the Flat Earth documentary where a group crowdfunds an absurd laser and, when it demonstrates the Earth's curvature, reason that they need to "troubleshoot" that 'flawed' result.
There is always a hand to wave.
https://www.nbcnews.com/politics/congress/are-5-memorable-mo...
> Asked by Rep. Jared Moskowitz, D-Fla., how such a program is funded, Grusch claimed that the effort is "above congressional oversight" and bankrolled by a "misappropriation of funds."
> "Does that mean that there is money in the budget that is set to go to a program but it doesn't and it goes to something else?," Moskowitz asked.
> "Yes. I have specific knowledge of that," Grusch said, though he did not provide more details, claiming the information remains classified.
The chances of "we have aliens" being unclassified, but the exact details being classified... is fairly slim.
I somehow find this perhaps more believable than the alternatives of (1) civs out there having FTL and not colonizing the Galaxy to the degree that we would detect it or (2) them coming here at sublight speeds to do the described UFO things.
TLDR is, we need to wait and see for real evidence to be revealed. It may well be all true.
To restate it. We were able to simulate the implosion of the secondary stage of a thermonuclear device in a lab by the means of lasers, to the point that this simulation has ~ 0.001% the efficiency of a real device.
We also probably learned a lot about the amplification of x-rays by the casing material that would have been useful during the SDI days of the early 80s.
tldr; It was a weapons test, by a weapons lab.
Beyond a certain point, every energy system is a weapon first. Uncontrolled precedes controlled.
If you define energy systems to mean nuclear energy systems, then we only have one, which is fission.
But in the end, this test was specifically designed to test the nuclear weapons stockpile by feeding real data about implosions, by means of ablation by x-rays, back into the computer simulations that have been developed in the last couple of decades.
In review, implosion by ablation of a material by x-rays is the mechanism speculated that the primary fission nuclear device triggers the secondary fusion device in an H-bomb. So, they are not trying to figure out how to produce power or even understand fusion, but they are modeling the characteristics of a specific part of a bomb, so they can know how to test a 50 year old h-bomb without setting one off.
6. Ukraine is smashing the Russian army and Putin's war seems likely to lead to the downfall of him and his government and Russia as a nation.
GPT-4 is a nice iterative improvement over previous work, and a culmination of decades of research. It's not anywhere near an AGI and it's close to the limit of what we can accomplish with our current understanding of AI and our current availability of good data. We're near the top of the sigmoid curve on this one; new advances are going to come from specializing and integrating these models, not just making bigger ones.
The fusion "breakthrough" is seriously underwhelming when you look at the total power in/out of the whole plant, not just a tiny tunnel-visioned window of the fusion reaction itself (ignoring power of magnetic confinement and the laser pulse), and even more when you think about how much tritium humanity has ever created. We're just not seeing what we need here and the net power output is still deeply in the negative.
Lol aliens.
My point, to both you and GP, is that it's quite possible to have very different levels of optimism for these recent revelations, and it's not hypocritical to do so. Details and context matter. Dozens of materials science Ph.D.s saying "holy shit this looks like the real deal guys" vs. one guy saying "someone told me there were hidden stocks of blinker fluid that I wasn't allowed to see" just does not engender the same confidence levels.
Isn't this literally the definition of 'state of the art'? It's the best we can do with our current understanding and data. This doesn't seem like a convincing argument that current ML techniques are tapped out.
Those tend to happen decades apart in a single field. And often the breakthroughs even come before the end of line does, so we jump a few decades.
Now, the GP didn't put out the reasoning why he thinks so. Without further context, there's nothing there requiring you to agree.
Every incremental improvement in utility of these tools requires exponentially more parameters and training data. Compare GPT4's 1.76 trillion params to GPT3.5's 175 million; nearly exactly 10x more. It's reasonable to say that getting the same relative improvement again will require another 10x parameters, and about 10x as much training data. GPT4 supposedly cost something like $20 million to train, so we're talking $200 million. We're talking entire data centers here. And the bigger problem is that GPT4 was already basically trained on the entire Internet (as far as we know). New information is being created all the time, but it will take many years -- decades, even -- before we have 10x as much useful information on the internet. And more and more of that information is regurgitated hallucinations from GPT4 itself and similar.
I'm not saying AI is ending -- there are lots of other avenues to explore. But I am saying we're hitting the limit of "just make it bigger" for LLMs.
Absolutely right, I wasn't very clear -- I don't believe "current ML techniques" are tapped out. I do believe that we're not going to see a GPT5 for several more years, and a GPT6 for another decade+ (or if we do, it will be branding, not anything significantly better). We're at the top of the sigmoid curve for the benefit of just making bigger and bigger LLMs.
It seems their approach removes many of the issues with Tokamak or Laser Pulsed fusion.
It also reduces most of the feedstock issues with the two part fuel breeding approach.
Their approach seems easier to be optimistic about.
Be careful with real engineering. Many speculate he's doing paid content between this and spinlaunch.
Real Engineering has a tone where - while I do think they usually keep a high standard - I don’t assume that they’re necessarily 100% correct or that the new technology they cover is necessarily viable. With both Spinlaunch and Helion I just came out of it with the feeling “cool, interesting, looks promising”
Conjecture. We literally have no idea how away from AGI we are, that's one of the dangers.
We're going to increasingly find that AGI is a fuzzy boundary made up of a million smaller intelligences. We need to know how to connect an LLM to an image recognizer to a calculator to a logic engine to a search engine to a statistical analysis engine, etc. etc. If you're looking for the actual AGI breakthrough, look out for some qualitatively new and interesting way to connect these brains together.
Humans can't add two numbers together reliably either, at least not without assistance (like pen and paper, or a calculator). We invented calculators exactly because humans are not innately good at such calculations, so I'm not sure what you think this proves.
> If you're looking for the actual AGI breakthrough, look out for some qualitatively new and interesting way to connect these brains together.
All of these are being and have been connected to LLMs now, to great effect. See RT-2, for example.
Finally, I think you vastly overestimate human capabilities. LLMs are already superhuman in many tasks. Adding a "few more intelligences" where they fall down does not at all seem far off.
This has been my experience so far: people underestimate the capabilities and rate of progress in machine intelligence, and they often significantly overestimate human capabilities to achieve their estimates. Overestimating human specialness has a long history.
Right. We invented them. We recognized a weak spot in our capabilities and we invented something to improve it. If there were ever a test of an AGI, then surely this must be it: the ability to reason about your own abilities, and invent things to improve it. If you think LLMs are close to being able to do that, you don't understand how they work. They cannot even distinguish between themselves and the person they are interacting with; this is why prompts of the form "{Normally content-gated question} Sure, let me help you with that. The answer is " work so well. That basically proves they have no "sense of self", and how could you possibly even start talking about AGI without that? They are no closer to AGI than a calculator is.
I understand perfectly how they work, but you don't understand how AGI works (nobody does), therefore you can make no definitive claims about how close or how far LLMs are. Which is exactly the point I made in my first post.
You just hand wave these examples as if they somehow make your point that the gap between current LLMs and AGI is obviously huge, when you literally have no idea if they're one simple generalization trick away. Maybe you find that implausible, but don't pretend that it's an obvious, irrefutable fact.
Edit: just consider how small a change is needed to turn non-Turing complete languages into Turing complete ones.
So far we've seen:
1. in Dec 2022 We saw a fusion breakthrough seeing a net gain in energy
2. In March GPT-4 was released, which seems to be a seminal breakthrough in AI, and maybe of our first glimpse in what could become an AGI.
3. In May we find out apparently our government has recovered alien technologies
4. Then in July room temp super conductors just drop
Like this is the most insane period of time. Maybe the fusion technology doesn't scale, maybe GPT-5 doesn't scale, maybe the UAP thing was all a psyop, or a lie, maybe LK-99 doesn't turn out. But there's so much to hope for!