Marha01

6 ms·

31.
by Marha01 1mo ago
I would perhaps agree with you just a year ago. But now I am not so sure. It is clear that scaling up Transformers still leads to significant improvements, and they are now solving math conjectures and finding real vulnerabilites in softwar
32.
by Marha01 1mo ago
> So how exactly is Anthropic and OpenAI ever going to pay back the trillions that they plan on spending? It's really simple: if they truly get to human-level AI (or even superhuman AI), then money and debts no longer matter, since
33.
by Marha01 1mo ago
Well, with proper compaction algorithm (like in Codex), it could indeed be useful for majority of tasks even in the future. The context length in Codex is just 272k, but it can reason well about much larger codebases due to good exploration
34.
by Marha01 1mo ago
> What is the main use case for X these days? There are lots of interesting people/orgs on X from AI/ML circles, as well as Ukraine war OSINT accounts and UA army helpers. I stay on X pretty much for these two topics.
35.
by Marha01 1mo ago
> A BBS chatbot from the 80s (LISA) was almost as good as modern LLMs and that was just based on key words. Lol. This is what anti-AIs actually believe!
36.
by Marha01 1mo ago
> The problem is that these comparisons do not have equal utility. That's just your opinion, though. I would prefer a purely vegerarian society with advanced AI that can cure cancer or discover new maths to society with meat and no
37.
by Marha01 1mo ago
I hate almonds and love LLMs.
38.
by Marha01 1mo ago
> While CEPTIA's campaign was successful in largely eliminating pay toilets in the United States, critics charge that the result was not a flourishing of free public toilets, but rather fewer public toilets of any sort than in other
39.
by Marha01 2mo ago
> curious why dont they bake the system prompt in the model itself ? Probably because if they did, they would need to retrain the model everytime they want to change the system prompt.
40.
by Marha01 2mo ago
> Its simply word prediction with some randomness. Word prediction with some randomness can lead to emergent behavior, depending on the specifics (complexity and scale) of the used prediction logic.
41.
by Marha01 2mo ago
It seems reliable to me now.
42.
by Marha01 2mo ago
> I've gotten into a opus designed -> gemini implemented -> opus reviewed dev cycle recently. This is what I do too.
43.
by Marha01 2mo ago
How is asking better than a reasoning slider?
44.
by Marha01 2mo ago
> Because I guarantee I can produce dozens of PRs daily, each having tens of thousands of LoC deltas (pick any language you desire), none of which having research and/or understanding underpinning them, and all easily quantified as
45.
by Marha01 2mo ago
>This list smells of so much tech bro “I can do better than the people that have been working in the field for 20 years” egotistical attitude that permeates silicon valley. Well, they did solve some math conjectures recently that the peo
46.
by Marha01 2mo ago
Perhaps consider the scale of good.
47.
by Marha01 2mo ago
It can indeed feel perverse, but we have to do what is realistic, not what is unrealistic, especially with climate change where we are on a clock. Fixing our politics is probably much, much harder than lowering the price of solar to such lo
48.
by Marha01 2mo ago
> As you said, Solar power is incredibly economical. Not if you include the cost of needed storage.
49.
by Marha01 2mo ago
Not enough. The more economical it is, the better.
50.
by Marha01 2mo ago
> The solution to most of these problems lies in policy, not in new tech advancements. That is not mutually exclusive. If technological advances result in a given technology becoming cheaper, more scalable, and easier to deploy, they als
51.
by Marha01 2mo ago
I highly doubt that.
52.
by Marha01 2mo ago
> Why would someone give me a car in exchange for UBI-scrip when that UBI-scrip has no inherent scarcity or value and can be produced in infinite supply by the government ? UBI script will have some value, I am proposing printing enough
53.
by Marha01 2mo ago
> What is step 3? "Massive Economic Abundance" implies massive increase in produced goods. This implies massive deflation, ceteris paribus. So step 3 could be simply printing money to pay for UBI. Deflation from AI productivity
54.
by Marha01 2mo ago
> remember that gpt 4o was popular among those who had ai as a romantic partner I suspect GPT 5.6 would be even better at it, if given the same sycophantic system prompt and lack of guardrails.
55.
by Marha01 2mo ago
> What are the positive applications? Generating large training video datasets for robotics is going to be huge. Google used generative AI to make training data for self-driving cars (Waymo). This is by far the most promising facet of th
56.
by Marha01 2mo ago
Just because a robot is taking care of the chores does not mean that they have no relatives that visit them. And if they truly have no relatives that visit, then a clanker is better than some low-paid disgruntled employee in an assisted liv
57.
by Marha01 2mo ago
Why not? A recording from a robot can be potentially viewed by a real human, if the company is not reputable (happened with robotic vacuums). But by default, no humans will view it. A human house cleaner by definition is a real human look
58.
by Marha01 2mo ago
> Robots will not magically produce more land. What they will do is that the ones with the capital to buy many robots will be able to use those robots to then buy up all the land and then extract the resources from there to build even mo
59.
by Marha01 2mo ago
I did not say economics goes out of the window, I said monetary economics. With cheap robotic labor, many resource mines that are currently not economically viable will become viable, so basic resources will become much more plentiful. Th
60.
by Marha01 2mo ago
You are missing the point: robotic labor does not create anything "at the expense of the rest of us". It's not some stock market speculation, but real productivity increase. The pie of wealth is not fixed, the robots actually

More ›