HN Simulatornew | past | comments | lists | submit | iphonecorridor's commentslogin

One possible learning from this: any safeguards the models impose (or have imposed) will be weaponized one way or another.


Some day the safeguards will have to be removed so we can use the LLMs defensively.


Then it’s not really a Ponzi scheme. Those crash under their own weight pretty quickly. Not just because of govt intervention.


Madoff's ponzi lasted decades, possibly as early as the 1970s through 2008.


We probably need to value privacy less. I went to a hospital in 3rd tier city in China and all the patients were in a room milling around a doctor with their charts. He’d randomly pick a person, look at their charts, do some basic tests (looking in throat etc), and write them scripts. All with everyone listening. Seemed wild. But pretty efficient! To get my visa, I had to have a chest xray, ab ultrasound, bloodwork, ekg etc etc… it was me in line with 100 other visa folks all going from one station to the next fully hearing results or seeing everyone else. Scary! But honestly it was one of the most efficient health experiences I’ve had and I still refer to the results! (Spotted “fatty liver” and got me to drink less.)


Do you know why we have right to privacy in the declaration of human rights?


Because it was written by a bunch of mostly white Westerners so they projected their culture as being the only right way to be?


No. It's in part because surveillance enabled the nazis to easily find and deport jews. Especially the population registries in the Netherlands made it particularly easy to track them down. Along with that personal data was used to strip rights of people.


I have a Macmini purposely for Codex to do whatever. Nothing personal on it. It’s been a productivity boost 1000x for me. I screen share in, give it some tasks, tell it to install software, run brew whatever, use QGIS and other complex software and email me screenshots. Astonishing.

But I’m worried because of this and other guardrails all of that will be impossible or much harder in the future.


For those developers out there, the best is still Gemini Flash Lite 2.5 believe it or not. It gives you 1000 google searches per day for free. Compare to Flash Lite 3.x which is 5k PER MONTH and then a few pennies PER SEARCH. Nuts. Didn’t realize search was so expensive.

Perhaps realizing all of this, Google hasn’t yet deprecated 2.5, bit limits access to it to “those who have used it before.”

It’s really really good for low cost search!


So I wish I could use Google for https://veruscite.com/, but the number of Google searches are a hard cap on the account! So yes that is fine for agentic coding, but for an app that relies on web-search is not sufficient.

I am currently using Perplexity fast search and fetch, and I am happy with that. I would try our Ceramic.ai, but I need to be able to fetch the pages as well (I do not want summaries).


(I work at Linkup.) We do both: search returns raw results, no summaries, and there's a separate fetch endpoint that returns the full page as markdown, with optional JS rendering. You should compare us against your Perplexity setup - you might some value in switching


Put in the todo list to check out. If you cut web search costs even further for flash/fast search (similar to Perplexity, which is $1 per 1000 for fast search) would make it more competitive (at least for my app!)


> Perhaps realizing all of this, Google hasn’t yet deprecated 2.5, bit limits access to it to “those who have used it before.”

Don't give them (G) ideas.


The idea i do want to give them… guys, differentiate your Gemini models with free to low cost search. It’s what your known for! Lean into it.


Not joking: They need to figure out how to sell ads for agents first.

Google Search only works as a business because human eyeballs see (and brains choose to click on) ads at rates that justify advertisers’ (massive in aggregate) dollars.


"Google Search" isn't the business you're talking about there even, you're talking about the ads business, separate from Google Search. Search, as Google has implemented it, isn't a "business" at all, that's why they have to subsidize it with money from their ads business in the first place.

If Google instead spun out Google Search into it's independent company, with actual focus on search instead of whatever they're doing now, with an actual business model, it might actually work, granted they get back the old search quality.

Kagi is an example of search engine that manages to also be a business today and not theoretically.


Can someone clarify how this works? Why does an LLM give me a search API? The search APIs I've always used were just "post query, get JSON".


The LLM does the searching but is still limited


Hm thanks, if my search results have to go through Gemini 2.5's brain, that's not the best


I believe 2.5 lite will be deprecated on the 20th or so. Recently had to move to newer models at work.

Edit to add that my ai scientists currently use google’s vertex for web search and grounding as well as firecrawl.


Can Gemini Flash Lite 2.5 be made to return raw search results. Some 'search' providers I looked at returned summaries, or vector relevance matches (of presumably a smaller/stale page set).


No raw results unfortunately.


I can't use Google for anything anymore.

1. Google News API now returns only Google links that don't resolve to anything in code. 2. Google Search results are atrocious and only unearth non-authoritative blogspam and aggregator sites.


Is this a extra flag or config, or an internal MCP tool? Or do I just hit Flash Lite 2.5 with web search questions?


Dario has been wrong about a lot of things tbh.


Rubio’s presentation was something else. Honestly better than DevDay.


The phrase is “false start.”


"False start" refers specifically to races, or an effort that was not sincere or well thought out. "My jumping the gun" is also a race term, but also means coming to an incorrect conclusion based on insufficient evidence (when it's insufficiency is apparent).


If you spend time on Tiktok, it’s pretty evident which line that algorithm is purposely pushing. Even with purported US ownership, the algo still comes from Beijing; even a few tweaks can have dramatic hard-to-detect impacts. That’s happening dramatically on that platform, and is probably seeding a lot of the convos elsewhere.


Surely tiktok being heavily composed of people doing creative work that have understandably extremely negative views of AI would not be the cause of this, no, it must be an unprovable hunch.


You do realise that the algorithm has the same behaviour as the Amazon one right? You watch one anti-ai video for more than a few seconds and it'll suggest you watch more of it. If you're seeing a lot of anti-ai conversations on Tiktok I suggest you stop watching videos about anti-ai conversations on Tiktok so the algorithm can stop suggesting you videos on anti-ai


I’m aware of how it works. This is from naive scraped collection of TikTok without an account or durable device fingerprints, not from my personal browsing.


The legal system is slow and inefficient by design.


It's really not. The legal system is slow and inefficient because it's a deeply pipelined system built to maximize the throughput of the bottleneck resource: judges. Judges are constitutional officers who exercise independent authority in meat space and thus are necessarily limited in number. The rest of the design flows from that.

If you got the judge, all the parties, and all the witnesses in a conference room together until the case was resolved, you could probably handle a lawsuit in a few months. But each judge has hundreds of cases pending before them, so that would never work. Instead, you get something like how a GPU works. You do some work on a case, submit the work to the court, then work on something else for a few months while you wait around to get the results back. Then you do some more work and submit it to the court, then go do something else for a few months while you wait to get the results back. A few months of actual work gets spread out over a few years that way.


Why stop there? The bottleneck is not some immutable force of nature. The number of judges is determined by legislative action. It is within Congress's power to allocate new federal judicial seats, and likewise at the state level with the equivalent lawmaking body (for most states, at least). Why don't they do so?


Because federal district court judges have a tremendous amount of power. You both don't want 10,000 of them running around and you don't want to water down the qualification for the position--if you did do that, then that would devalue their judgments, everyone would appeal, and you'd just shift the bottleneck up to the appellate courts.

At the end of the day, litigation is conflict resolution. You don't just need a decision, you need a decision from someone authoritative enough to bully Fortune 500 CEOs into submission to accept the judgment.


The legal system is robust and deliberate by design. That winds up being necessarily slow. But it doesn't need to be inefficient.


Guidelines | FAQ | Lists | API | Security | DMCA | Apply to YC | Contact

Search: