HN user
H+
"Early in the rollout of the technology, employees also faced restrictions on using Gemini to write or analyze software over concerns that proprietary code could leak into the AI model’s training data, they said"
Wait, what?
If in 2020 I had sent you a book about the LLM achievements of 2026, you would probably have thought it was a science fiction book with no relation to reality, wouldn't you?
Perhaps it will. And perhaps countless animals are already better than us at appreciating a good sunset, yet we do not seem to value them much.
Personally I am happy that there are weirdos and that they share their weirdness.
What he has may not be a very serious disease yet I was interested in learning about his experience and felt sorry for him.
Did he somehow wronged you or are you just having a bad day?
Because supposedly they would repair themselves or be repaired by other robots, and energy would cost less and less, anything would cost less and less if work is increasingly done by robots than can be improved year after year.
Would it be more scientific to say it is sure it will happen?
I suspect the opposite is happening, too many times an environmental catastrophe has been predicted in too certain terms and has not happened, which is why many people lost trust.
And if you think that we shouldn't try to predict but only inform about what has already happened that seems even stranger to me.
But this paper is not about extra EU migrants but all migrants. And even then if we control for age they say they are contributing less than natives.
I think it would be very odd that less educated people on average contribute more than natives, especially if they are at risk of being discriminated when looking for a job.
Is the SpaceX 1.7 trillion IPO on Friday some kind of psychological anchoring?
Say it goes down 50%, then it suddenly appears cheap, when in reality that would still be way overpriced.
But surely Google has both ML people and people expert at optimising stuff, be it hardware or software. In my opinion they have the talent, the sheer number of employees and the capital. Can deepseek really have people much more talented at optimizing stuff?
But what is preventing their competitors, who have many more employees, who are also very talented, to do the same?
Every little improvement would save them billions, so it's hard to imagine they aren't pouring a lot of resources into that already.
"Furthermore, vegetarianism, though morally laudable, has an obvious economic limitation — when one person refuses to eat meat, it lowers the price of meat for everyone else"
I very much doubt that, I think the opposite is happening in the long term because of economies of scale.
So go ahead, become vegan! You already know you should!
They talk about subsidies, "draining its coffers", and "taxpayers helping some of the richest corporations on the planet buy servers, equipment, and power infrastructure", but it doesn't seem like they have lost any money, just that the tax breaks mean that they haven't earned as much as they could.
Not that I think that those datacenters should have those tax breaks, but the language seems quite misleading if not an outright lie, presumably without those the datacenters would have been built elsewhere.
For the third time I never said it was a good plan.
Yet LLMs can play chess and have a "mental" representation of the chessboard.
If LLMs get better but do not progress at playing games when not specifically trained on it it seems to point to a generalisation failure, a limitation that would prevent LLMs to ever achieve AGI, I do not know if that is weird but it seems that for now nobody really knows if they can achieve AGI or not. Perhaps some emergent behavior will arise after more scaling.
To me it's only totally unsurprising if you are 100% certain that LLMs will never reach AGI (like LeCun thinks for example).
Not saying this is a smart plan, but how is it far more stupid than invading Iran which is basically impossible unless you are ready for tens or hundred of thousands of casualties among your own citizens?
I suspect the plan was (and still is) to weaken authorities in Iran so that the people take over. Or have the Iranian government reach a deal that would be less favorable to them.
A plan with quite long odds you could rightly say, but not as stupid as subduing them by invading them I suppose.
"Investors who have poured hundreds of billions into closed-source labs are betting on an unprovable safety moat".
Nobody is investing in closed-source labs for safety reasons, being able to explore more in details what and how the model is thinking is nice but by no means a game changer. What matters to investors and most of the users is that the model gives the right answer at the end.
I do not think there is anything wrong with that (just that this is uncommon and surprising, so I'm curious), but in my hypothetical scenario where storage isn't an issue, having useless data stored should not matter.
Changing the default is not a big deal, for sure, but would you still do it if it required 10 minutes of your time every 6 months? If yes, I just cannot believe that you have no reason to do it. Perhaps it makes you feel freer, perhaps you think your device is more organized or cleaner, or something else.
But of course in real life data uses storage, but I feel like most people are on the other extreme and keep too much (data or possessions) because they cannot let go, but chat data footprint is usually minimal unless you receive lots of photos/videos.
But if storage isn't an issue, why go out of your way to remove them? Don't you think there is an underlying reason you are changing the default to remove them?
Why do you feel the need to delete your chats? Is it to save space? Would you still do it if you had unlimited storage?
Also it keeps asking you for execution permission all the time for the same commands over and over again (even if you add them to the settings).
Worse, I selected "Terminal Command Auto Execution: Proceed in Sandbox", and it keeps switching to "Always Proceed" (with a nice warning about how it is very dangerous). I have changed it 10 times then just gave up and switched to Codex.
They announced Gemini 3.5, an AI centered search approach, Gemini Spark, Gemini Omni, smart glasses, and more, and somehow that accounts to "having nothing to say"?
How high is the bar? Sure, most of this stuff is just improvement on existing things, and it's true that if you are not interested in AI then there wasn't something for you there, but are we expecting them to announce a revolutionary product each I/O conference?
How do they calculate that?
3.1 has 57M output tokens from Intelligence Index, 3.5 Flash has 73M, so not a lot more, and 3.5 is a bit cheaper, I don't get how 3.5 can be 74% more expensive.
Couldn't get it to work on Chrome. Had to find the stream on youtube https://www.youtube.com/watch?v=wYSncx9zLIU
The end goal of these companies is AGI, or even ASI. If you believe this is around the corner, and think AI can do the job of a human for less money, it makes sense to put all your money into working towards that goal and buying as much compute as you can. This is especially true since whoever gets there first (or is simply ahead and can use their AI to get even better) gets a big advantage.
That wouldn't help in that case as exfiltrated data is committed to public GitHub repositories. Unless you have to accept every time an app posts or requests data from known hosts?
Ok so the hype would be people saying AI can currently do something well and autonomously when it cannot (or not consistently enough), and it is easy to prove them wrong.
But I feel like people are more hyped about what the AI will be able to do soon rather than what it can do now.
I think AI does understand things (depending on your definition), how else could we communicate and ask it a question if it didn't? I mean we're quite far from Eliza here.
And yes, often their answer would be so wrong that we think it is impossible that AI understands anything, but this jagged intelligence doesn't prove, at least to me, that there isn't some understanding. At what point do we say that AI understands things? What if we can reduce 99% of those dumb failures, would we then say than AI understands?