I think bloppe was talking about the Tulip hype cycle that took out many generations.
90 karma · joined April 27, 2024
The unemployment figures is mostly a political projection, like CPI.
In 2026 everything fron sewage to chickens to oil to education is now tied to AI. The entire economy is now, for good or bad tied to AI.
What most people either didn't notice or understand is that the U.S. was entering a recession and would have been in one if it hadn't been AI
Your Railroads and the Internet parallels are so very apt. We certainly are very productive even in 2026 off those but almost no one remembers how many large behemoths were wiped out in the U.S. because of them.
Yes, it does seem to me that the AI of 2026 will be like the Railroads and the Internet - fundamentally transformative technology that changes how things are done into the future but there will be blood in the short term, which is why I would not be participating in the IPOs directly
Arguments are being made that, with the midterms over and 2028 just around the corner, a less sympathetic administration would create significant headwinds, adjusting the valuation of Anthropic and OpenAI down.
What we don't know are what Anthropic and OpenAI have planned and in the works.
A large population of extremely skilled and talented people are being laid off by their employers to make space for AI investments. It's likely a significant number of them will start their own business with AI.
AI is also gaining a much wider audience and acceptance. It's likely over the next few years, AI demand will grow, not stagnate.
That said, IPOs are usually liquidity events for LPs and for something really disruptive and unpredictable as AI, I won't be engaging in their IPOs. Only time will tell if this was a good decision, or bad.
- provide housing, education and medical coverage to people who are trying their best to make a living
- support a person who has killed atleast one more person for reasons that a jury of 12 people found unreasonable
Money unfortunately is zero sum.
In the U.S. the state has never been an arbiter of life and death. A jury of peers decide that.
In a state where we struggle to house people, home prices are exploding through the roof, who bears the costs to hold those prisoners for the rest of their lives? to feed them, to provide them a habitable bed, cleaning supplies, medical health as they age and start to need increasingly more care?
and the costs to hold those prisoners for the rest of their lives? to feed them, to provide them a habitable bed, cleaning supplies, medical health as they age and start to need more care?
absolutely fair criticism. I think this is how it will work out for students (my specific comment is written for devs but the underlying concept applies): https://news.ycombinator.com/item?id=50361121
To repeat what I said here: https://news.ycombinator.com/item?id=50370496
What would be fantastic if we, as a society, invested in (S)LLM tools that every student "owns" and can "tune" to their liking. Almost like an assistant that "grows" with them.
The idea is that these tools are theirs, in their control and for them to manage.
Unlike Fable, it doesn't arbitrarily stop a student interested in reversing software or hardware from learning the craft because they are not a licensed researcher.
> But there's a difference between AI helping the teacher deliver a tailored learning experience to an individual, and a student just getting answers from AI. They're totally different. It's the second one that we want to simply block
What about an entirely different approach to that?
Existing study plans weren't designed with generative AI in mind at all. Educational policies and study plans even go back as much as 2000 years - we still follow practices and principles that were used when there were no electricity, forget PCs and generative AI.
Young children used to flock around a knowledgeable elder who would impart knowledge under a tree or a covered shed.
Maybe we need to rethink those policies, study plans, practices and principles by asking ourselves "what would the next generation of education look like if all exams were open book"?
What are some uses for which you found Luna to be unusable garbage?
From my side, I use Harbor to score tasks and Luna certainly performs when the prompt is either well or overspecified.
And this is exactly the argument I make on this thread. You might be unable to see it though because apparently it's a deeply unpopular take.
I made a main comment to this very post with a similar perspective that's downvoted so heavily I can't even see it while logged out.
Is there a possibility that in that productive struggle, certain kids just give up and decide either education as a whole isn't for them, or a certain subject isn't for them?
Is that a good thing or a bad thing for the kid?
Is that a good thing or a bad thing for society?
Could we have an equally productive struggle while fully integrating Gemini in all its glory? If yes, how would that look like?
It also seems you are overindexing on current, volatile events. Unless you're in Quantum space, Newton's laws haven't changed in a while. Sorting algorithms still function the same as does the x86 instruction set or how the human body reacts to ibuprofen.
Did I get your point, or am I missing something?
I encourage you to try and access the internet through a school's firewall. If you think Fable restricts you from looking controversial topics up, a typical public school's firewall is likely even more restrictive than the Great Firewall of China.
Honestly it's too early to know.
Also, schools are in the business of imparting knowledge and schools in the U.S. are extremely expensive. Any parent knows the additional fees, supplies etc most schools demand on top of their taxes that go to the school district.
If you take the daycare aspect out of it, a $500/mo ChatGPT subscription is noise by comparison. A student doesn't have to be lucky to have a receptive teacher that has enough office hours and the interest to help the student. ChatGPT is available 24x7. A teacher takes days off, falls sick and has favorites.
If AI enters the school with full force, bulk of the teachers are likely going to be replaced by various care staff.
There will still likely be some human assistants around who will be the new "teachers" but the current format of everyone staring at a whiteboard is going to change forever in this future.
The concept of "grades" (as in K-12) are also likely going to go away replaced by cohorts, simply because every student has their own "teacher" who understands their specific needs and can pay them all the attention they need assisted by a human adult.
One child can be in one cohort for English and an entirely different cohort for Math.
that wasn't my point at all. The point I was making is these tools make forever renters of intelligence. That's why I mentioned the Black Mirror episode.
It's not enough to have options in which service to rent forever.
Imagine if you had to rent your own own dictionary or calculator forever.
Not entirely accurate. Public schools might be consistently 20 years behind on things but Private schools certainly are usually pursuing some form of bleeding edge. That is in part why parents send to private schools.
That is by design: Public schools don't get to pick and choose who attends and thus don't get to pick and choose who the parents are and thus, what the environment at home is like.
15 years ago WikiPedia was extremely controversial. They were full of unverified articles, claims and agendas - the 2000's equivalent of "hallucinations".
As WikiPedia has gained inertia, popular articles have gained more eyeballs, more attention where those issues are mostly gone.
In 2026, we haven't yet figured out how to bulid LLMs that will safely tutor students without also telling them how not to make their own drugs or to kill themselves because Dad always screams at them when they get home and they want it to stop.
Even if an LLM had a 1% error rate, in Public schools that have thousands of students per county, someone is going to die from what an LLM said and that one lawsuit is going to bankrupt that school.
Not really. Most services and labor isn't elastic. For one, you cannot service more than a dozen people a day if each service takes 30m. To scale, you will then need to hire more people who will want to work on a salary and will demand healthcare and days off.
> The lead that the landscaper in your example is building now actually compounds over time, regardless of whether competitors adopt grok bot in the future or not?
You are conceptually right except that you are conflating cost of service with a business relationship.
If my business relationship offers me the same cost of service for the same service as their lower priced competition does, it's then not even a thing for me to worry about. They are completely financially fungible (plus the value of the business relationship).
Assume I work with landscaper A while landscaper B is using grokbot to be more efficient.
If my business relationship with landscaper A is strong, I will happen to mention to them what landscaper B is doing and if landscaper A values our relationship, they will try their best to do so as well or comp my service accordingly.
There are other factors like churn, overhead, regulation, licensing etc that I am happy to discuss in more detail.
Look at the graveyard of ChatGPT wrappers. Same concept.
This is why a grandparent observed that amortized ROI that depends on a 3rd party service becomes 0 - in capitalism, an arbitrage opportunity is competed away.
This is why companies build their own in-house talent for their core competencies - so that their ROI on their core competencies doesn't become 0.
This is also why sama and Dario are trying to regulate OSS models - because those are a direct attack to their core competency and if left unchecked would cause headwinds to their ROI,
For back office operations, one calculates ROI over 5 years or so as upfront investments are rather large (less for turnkey SaaS but that's a whole another discussion and not what's happening here. Here the owner is building their own solution).
Your theory is sound. What you're missing is in a rational market, 3rd party edge only provides an initial boost until the rest catch up. That's why companies invest in in-house talent, so they own and control it.
If you're in the U.S., what's stopping you from doing that? Some of my coworkers in tech retired to the country, raise their own chickens, goats, grow their own food and are all on solar. They seem to be loving their life.
Yes it is. Pulling a simple permit takes months and thousands of dollars. Building a new house is practically out of the question.
The SF Bay is deeply supply constrained but just because tech billionaires here can buy up a dozen homes while offering millions over asking, the city gets their taxes and so they are not incentivized to improve things.
NSA doesn't have to post train Fable. Maven works well enough and that same infra. can be used for Fable. I imagine that's what's going to happen in 2 years once the current admin. moves out.
The next iteration of AI assisted devs will not have all the devs use the same harness and model like most do now.
Instead, devs will build their own post trained models that are experts of their own domain, that have their secret sauce..
I imagine any dev would want most of those models to be something they can self host because of multiple reasons - one of them being that their employers or customers might not agree to use a cloud hosted or managed model. Plus, costs. And control.
Post training is in the future because we haven't yet reached a shared agreement on what that first "good enough" model will be like - and also because the current AI scape is in a state of flux where post training will have a lot of churn.