vjvjvjvjghv

"Then they'll turn around and say there's a decent chance ASI will kill us all, so better we build it before China does."

And also set up moats so for security only they are allowed to push the state of the art forward.

"Right but why would VCs want this? Wouldn't VCs value the liquidity of being able to flexibly enter and exit positions? I really wish there were a simple explanation of why companies don't want to IPO as early as they used to but there isn't."

To me it's pretty clear. The earlier investors take out all the possible upside and then dump it on the retail market. The retail investor will be less and less able to buy stock that will grow 10x and more. It's hard to imagine SpaceX, OpenAI or Anthropic will multiply in value after the IPO. They are priced in a way that they have to execute perfectly for a long time to justify even their current price.

Just another step for the super wealthy to keep profits for themselves instead of letting the broader population to benefit.

I think we will see the same trend as other industries. “Good enough” but produced cheap is better business than “really good” but expensive. I assume most software will go that way, or has gone that way already. AI will probably cover most of it and only a few senior engineers will be needed. Basically like any other industry where it went from everybody being a craftsman to a few people building the machines that then can be used by relatively untrained people.

MS was famous for that. Seems since around .NET they have forgotten about it. Delivering and promoting toolkits to devs they don't use themselves.

Nothing. And it will happen big time once the AI companies become enshittified. Anthropic and OpenAI won't be able to produce the results that would justify a trillion dollar IPO without doing shady stuff.

Maybe it's better than deciding based on TV ads.

Wow, this feels almost like the real thing.

Is there something similar for the web now? Easy UI editor but a fully featured language behind with direct database access.

Are there any realistic ways to achieve AI safety? Whatever that even means. How can they avoid users doing stupid/dangerous stuff with the AI?

Is there realistically a chance that we won't get total surveillance in the next decades? Unless something dramatically changes, I just don't see how it won't end any other way.

I treat AI like an intern or junior team member. With enough guidance, they can contribute a lot but you can't let them loose without supervision or they usually will produce a big mess. As of now, you are still responsible for overall architecture. One strong indicator that something is going wrong are big pull requests where the AI has rewritten large sections of the code.

Does this mean they will run everything through an LLM? Besides the cost, does this mean the whole business will become non-deterministic? I can see using LLM to produce code that the business runs on (and can be tested), but running the business directly on an LLM seems to be asking for massive trouble.

I definitely feel AI data centers should mainly be built next to the houses of the big tech CEOs. I don't know prevalent this is, but I have seen videos of people living next data centers and the noise was just unbearable. It's also not ok that utility prices are rising because of data centers.

Exactly. It's hard to get excited about a technology that in theory could provide huge benefits for all, but because of our economic system there is a good chance that the majority of the benefits go to a few and many people will actually be losers.

I am believing more that with advances in AI and then also robotics the capitalist system as we know it will break. The question is what comes next.

"cheerleads when one party does it and castigates when the other does"

I always ask partisan people "what would you think about this issue if the other guys were in power". It's pretty sad how it has become normalized to find the same thing ok when one party is in power and being really angry about it when the other party is in power. Unfortunately Krugman can't resist that temptation either.

I am not educated enough to criticize what was done during COVID and whether inflation could have been avoided. But the way Krugman tried to explain away the effects of inflation in the NYT almost every week wasn't right. The whole "vibecession" thing was extremely condescending from a multimillionaire like Krugman towards lower incomes and probably one of the reasons why the democrats lost.

"and you probably have more reasons than what you're letting on."

Can you enlighten me on these reasons?

I would be more willing to listen to Krugman if he hadn’t spent so much effort during the Biden administration to find reasons why inflation wasn’t bad but just a “vibecession”.

The only responsibility these guys are feeling is to make trillions. Billions is not enough anymore. Maybe Meta is the most honest of them because they don't even pretend.

I know people who have vending machines. It's a very nice business but like almost everything else you need experience to do it right. It's not easy to find good locations, you need to stock the right stuff, you need to deal with repairs. It's not as hands off as you may think.

There is definitely much more happening than what american media tells us. When I visit family in Germany they are all doing pretty well. There are problems with expensive housing but that seems to be a worldwide problem.

But there definitely seems to be a lack of ambition in business and technology. For example, the car companies seem to be more coasting on past successes vs pushing things forward. Europe is also not trying much to compete in AI.

The real scandal is the prices on the menu she posted.

What do they test? In general, I don't understand the whole "AI safety" stuff. What are the scenarios we need to be protected from and how is this going to be accomplished?

"I think chat is a great entry point for many experiences. The approachability and flexibility are unmatched."

Just wait a few months/years and then some smartasses will come with a novel approach nobody ever has thought of before. Let's call it "context menus". They will give rapid access to the most used features without the tedious typing.

Until MacOS 27 the text context menus had items "Writing Tools" and "Proofread". Quick and efficient. MacOS has dropped these items and replaced with "Ask Siri". Now I have to type "proofread" every time and pray Siri agrees that it should proofread the selected text. Once it has done that, do i have to type "replace"? I don't even know.

There is a reason why UIs became popular. Command lines are fine but hard to discover and tedious to use. With AI it's even worse. You can't know for sure the AI interprets your commands in the way you intended.

The pause is probably rehearsed. I don't think anything Jobs did in public wasn't calculated.

My mom is older than any reactor that’s used for power generation.

France doesn't have an answer for decommissioning costs.

"California has those services but they tend to be only funded to a fraction of the need."

Sometimes it feels like they are well funded but the people running the programs live in a (very expensive) dream world that has no connection to the actual need.

" even a trillion dollar company, "

Exactly, not "even", these huge companies have to squeeze money out of everything. To keep their growth rates up, they have to make bigger and bigger swings.

" less innovation"

"Innovation" as in marketing expenses and stock buybacks?

I am good with getting rid of B).

A) can probably only be fixed with regulation. Why would hospitals and insurance otherwise make prices transparent? The current state serves them really well.