They can't even reliably follow instructions from text. I think "it's just around the corner/just wait x months/just wait and see bro" is one of the most telling signs of AI psychosis.
You do know that this was the same thing people said about crypto, right? And that the internet of things where your fridge connects to the Internet is hated by most consumers and had nowhere near the impact that IoT evangelists said it would?
This is such a creepy dystopian thing to say. Don't you realize that? Isn't this "yes, there will be pain, but the future is inevitable and we must go forward into it" attitude straight out of multiple horror and sci-fi stories?
Demos are also often misleading and cherry-picked. Using AI to do one cool demo that breaks down 99% of the time when circumstances slightly change has played an outsized part in most of the AI insanity we are living with.
I'm amazed you think that instead of using an LLM that someone will go buy a book and spend a week learning something that, judging by the fact that they last used it 30 years ago, likely won't be relevant for them soon.
I mean, yes, "sophisticated" people, including institutional investors, do fall for scams. See SBF, Theranos. These scams are often most effective when there's mania and FOMO about a technology that everyone seems to be saying is amazing and will make them rich, in spite of the current evidence, which sounds a lot like AI.
It's not even clear what this means. If an agent generates some code and I delete half and rewrite it, is the code AI-generated? If I start to write def calculate_average() and the "AI" auto complete fills it in when I hit tab, does that mean AI generated the code? What about auto complete which existed before LLMs? Did anyone ever measure how much code was "generated" using that?
It's really rare that you get someone claiming 10-50x productivity gains who posts some proof. It's not surprising that the person posting it has "contributed" nothing remotely of value except for a huge number of PRs that were closed as AI-generated spam and has been banned from contributing to multiple projects. These are the people telling you that you should be 50 times more productive and if you're not then you're "doing it wrong." You are crypto bro 2.0.
I know the bricklaying metaphor gets overused, but saying coding is solved and then seeing what kind of code ends up in CC or OpenClaw seems to me like building a retaining wall out of oddly shaped broken brick parts and wood and stones and saying building walls is solved. Technically it's a wall and can do what a wall does, so maybe who cares what went into it, but I wouldn't ever use it to keep earth from falling in and crushing my house. I'd hire experienced engineers and craftspeople.
I agree we can/should decide collectively. But when the vast majority of Americans have a place to live, air conditioning and heat, a refrigerator, enough to eat, etc. it's hard not to say that they're comfortable. As for fulfilling, that's describing a utopia.
I recognize there are still people who are food insecure, and I think it's an abomination how people are suffering (many poor rural black people in the south don't even have access to basic plumbing and end up with diseases like hookworm), but making wealthy people into a boogeyman just seems like an emotional argument so much of the time.
How do you run a society based on who is dissatisfied? It seems reasonable to say that, even though there's a massive wealth gap, if the poorest are healthier, wealthier, and generally better off than they were a generation ago (that might not be true though with this current generation, who it seems may die earlier than their parents did), then changing laws because those people are "dissatisfied" seems kind of arbitrary, because dissatisfaction is kind of the human condition.
How about the top 10,000? How about any wealth you personally have? At what point does it stop being OK for you to take people's wealth? When someone becomes a billionaire?
This reminds me of when Bernie Sanders was asked why it's OK for him to be a millionaire and he said something akin to "If you write a popular book like I did, you can be a millionaire too." Well maybe if you reinvent electric cars and space travel or build a company that ships you almost anything you can think of within 2 days then you can be a billionaire.
Wtf? A sub-agent is a tool you give an agent and say "If you need to analyze logs delegate to the logs_viewer agent" so that the context window doesn't fill up with hundreds of thousands of tokens unnecessarily. In what universe do you live in where that mechanism somehow means you need fewer humans?
Do you think this means "Build a car" can be accomplished just because an LLM can send a prompt to another LLM who reports back a response?
You're hearing a lot about the jobs problem largely because the media's job is to scaremonger. There have been multiple studies that have concluded AI's impact on jobs/layoffs is negligible: https://budgetlab.yale.edu/research/evaluating-impact-ai-lab...
This has become some kind of truism that has little actual evidence. It was said it would happen in 2022, 2023, 2024, 2025, and is still being said but it isn't true. The models got better this year but they've always gotten better. And none of the recent improvements have been as dramatic as say GPT3 to GPT4. I feel crazy. People are saying the sky is falling but every researcher or person who analyzes these issues professionally says the opposite.
Maybe this is obvious, but if my name were attached to well-known public OSS code, I'd definitely put forward more effort to ensure it's of as high quality as possible. Within a company you can write somewhat crappy code and have your close team members review it, and honestly there isn't much at stake as long as it works and you haven't done anything egregious. But yeah, you don't get promoted for following the style guide.
The hypocrisy was most notable in experts who said those protesting against the lockdowns (outside), who were considered right wing, were risking spreading the disease, but then said the opposite when the protests supported a left-wing narrative.
Also the CDC who said you had to stay six feet apart even outside who then were OK with people gathering close together during protests and shouting (specifically called out by the CDC as a risky behavior).
Google was calling itself an "AI-first" company beginning in 2016 or 2017. They designed and built TPUs nearly a decade ago and were using transformer models in products like Google Translate but didn't make a big fuss about it, it just made the product way better. People should at least credit Sundar somewhat for this, it turned out to be quite prescient, especially the advantage of having your own chips that are specifically designed for ML.
Some people also call evaluations "tests". There are unexpected things that come along with new models, like the model in a workflow you'd set up suddenly starts calling a tool and never stops or decides to no longer call a particular tool, so running your existing evaluations to catch regressions like this and potentially updating the prompts is considered "testing" your prompts and harnesses.