T430s as a mobile linux workstation for garage projects. For anything else, 14” M series macbook.
Thought I’d put Asahi Linux on it but the workflow I’ve settled on is to use my Mac as a thin client for a linux box under my desk. Best of both worlds.
With your experience in this space, are you able to provide some pointers for how to get involved in a productive manner?
The housing problem feels so daunting that I struggle to understand how one could help move the needle. Knowing which organizations to volunteer with or places to donate that actually improve the situation would be a helpful start.
When I lived in downtown Los Angeles, I volunteered at the Midnight Mission and various veterans organizations dealing with addiction and homelessness.
Only you can really define what is productive to you, but people are always needed for running the shelters, working in the kitchens and more. It isn't glamorous at all.
People need to be transported from one facility to another, to medical appointments, to court, to counseling and more.
I currently work with youth at risk, helping them finish their high school degrees and families who are right at the brink of falling into serious problems.
I can't really tell you which organizations to volunteer, as there is a spectrum of them; some are religious like the Salvation Army and some have social and political agendas that may not align with you. Without recommending or condemning, for example, you will find organizations who work with pregnant teens who are anti-abortion. This may not be for you. The same goes for sexual and gender identity issues.
One thing that seems very unpolitical to me, and it is how I started, was volunteering at a local shelter. If you have more affinity with helping women, there are shelters that deal with them. If you have more affinity with dealing with people with addiction issues, there are shelters dealing with them.
Because of my background in the military, education and addiction, I find affinity working with veterans, addicts and at risk youth.
The problem is daunting, and the needle feels unmovable most of the time. In the big scheme of things I feel powerless, but I know reducing local misery is possible and is something I can help with.
> What if it can be Einstein, but can’t draw a Pelican, write a solid college-level essay, or fold clothes?
Then it’s an expert system.
Stephen Hawking wasn’t very good at folding clothes.
The ‘General’ part of the term ‘AGI’ seems like a trap to me, because there will always be new workflows to master. Can Astra one-shot level completion on some yet-to-be-released video game? If no, does that mean it’s not yet ‘Generally’ intelligent?
You won’t get pure ‘general’ intelligence until you find Einstein’s hidden variables and load the state of the entire universe into context.
Meanwhile, building a series of expert systems targeting specific valuable workflows is useful today and seems like it’ll continue to scale to cover huge swathes of economically valuable workflows.
I think that’s the more interesting thing to be measuring. The surface area of useful economic workflows that can be addressed with expert systems built with today’s tech.
Hitting some ‘Artificial Expert Intelligence’ coverage threshold on economically valuable workflows is what will matter for humans well before pure ‘general’ intelligence.
The only important part of 'general' is the ability to learn from experiential data and update your own model. That's what leads to general capability. Humans can't oneshot any task natively, but we can practice for a while until we uncover often novel methods of accomplishing something.
Therefore: the current transformer architecture is fundamentally incapable of AGI because the models have no mutable long-term memory.
You only have weights (large immutable memory), or context (small mutable memory).
Humans have mutable long-term memory: I can learn a new skill, adapt an old skill to new information, or learn new knowledge today that I couldn't perform/didn't know yesterday. I don't have a training cutoff.
Context engineering is an attempt to paper over this limitation. You can get really far with context engineering and huge models, but you will never get to AGI because there are many tasks where humans' mutable long-term memory outperforms.
For example, a human can invent a new musical instrument and then learn how to play the instrument they just invented. That's inference (inventing an instrument) leading to training (neuroplasticity). Humans have the ability to train our NNs with considerably fewer training samples. Everything that you can do with transformers is in one causal direction: training -> inference.
So if we take a huge with enough compute (CPUs, b200s, petabytes of SSDs), we install on it both the Astra, and the toolsuite to incorporate new sensory inputs (threads/sessions), camera, microphone, temp sensors, the lot, into a new version of the model. This model is then swapped for the old model, or traffic slowly brought over, or even adjusting weights in place.
Then my hypothesis is that thing as a whole could achieve AGI.
This feels like a very close approximation on how we humans evolve our brain. By encountering new experiences/sensations, classifying them as negative or positive to us, filling it away in neurons. Or by training motor skills etc. In the end we get more connections between neurons in our brain and we are capable of more.
Bingo, LLM architecture just does not lend itself to becoming AGI. They can get really good, sure, but they will always struggle with novel input and scenarios.
The more training data that is shoved in to them, the more they'll seem to solve novel situations, but in reality it'll be things that exist in the training data.
Aka Star Trek hologram characters aren't sentient, and actually anyone who things droids in Star Wars can think of a weirdo. C3-PO just kept running out of context and trying to revert to it's system prompt.
> Can Astra one-shot level completion on some yet-to-be-released video game? If no, does that mean it’s not yet ‘Generally’ intelligent?
If a model can't learn on their own to play some new game just as well as humans do, it's not AGI.
It's okay if they would take some hours or days of learning (like humans might), but if they can't do it at all during their normal operation, that's not general intelligence
> You won’t get pure ‘general’ intelligence until you find Einstein’s hidden variables and load the state of the entire universe into context.
But humans have general intelligence. AGI is about matching human ability, and we know this is possible in principle because brains exist
Creating “autonomous weapon systems that can kill people without human involvement” is not the whole point of Anduril. That is against SROE which are controlled by the DoW and it’d be a monumental ethical dilemma if this were to be changed.
Also, Palmer is one of many Anduril co-founders, and is not the acting CEO. His opinions are an input but are not ground truth for the company. He has effectively been a marketing figurehead for years.
Please stop spreading disinformation. If people engaged with defense technology and ethics in defense with more earnest curiosity, we’d have more productive dialogue and a less polarized tech community.
> Creating “autonomous weapon systems that can kill people without human involvement” is not the whole point of Anduril. That is against SROE which are controlled by the DoW and it’d be a monumental ethical dilemma if this were to be changed.
> Right now, there are so many weapon systems that require manning. You know, if I can have one guy command and controlling 100 aircraft, that's a lot easier than having to have a pilot in every single one. And it puts a lot fewer American lives at risk.
> To be clear, "autonomy" does not mean remote controlled.
> Once an autonomous weapon is programmed and given a task, it can use artificial intelligence for surveillance or to identify, select and engage targets. No operator needed.
> If you were having to require the human operator to actually map every single action and say, "Hey, do this. If that, then this." It would take so long to manage it that you would be better off just remotely piloting it. It's the AI onboard all these weapons that makes it possible to make it so easy.
To me that seems like he's describing a system where you give it a task like "kill all the insurgents in X area" and it'll go and do it without needing an operator.
> Also, Palmer is one of many Anduril co-founders, and is not the acting CEO. His opinions are an input but are not ground truth for the company. He has effectively been a marketing figurehead for years.
If the company's really good at heart, I'm not sure they would have kept a guy who keeps telling the media that the company is working on autonomous weapons to kill people without requiring an operator, and that he's frustrated that he can't make what he REALLY wants to because of all the pesky war crime treaties.
Thought I’d put Asahi Linux on it but the workflow I’ve settled on is to use my Mac as a thin client for a linux box under my desk. Best of both worlds.
reply