Hacker Newsnew | past | comments | ask | show | jobs | submit | evrenesat's commentslogin

WYSIWYG-IYL (if you're lucky)

I think Gemini 3 pro (high) in Antigravity does something like that because I can keep asking for different changes in the same chat without needing to create a new session.

I'd be willing to pay around $100 for a rechargeable version with a battery life of around 24 hours and 2–3 minutes of usage. However, a single-use battery would only be acceptable to me at a much lower price point, such as $30–$40.

China needs to build the world's trust and respect, while the US is slowly but surely losing theirs.


slowly?


“A user-agent is software that performs communication or interaction with another system on behalf of the user, historically stemming from earlier messaging systems, and literally meaning 'the user’s representative actor'.”

I guess everyone would agree that web browsers would come to our mind when we hear "user-agent" in general context. After years of forward and backward evolution of web extensions, AI powered web browsers inevitably will move forward the meaning of the term "user-agent" and deliver the value it actually promises.

Together with the advance of "browser-use" optimized local LLMs, current M4 powered Macbook Air can provide enough juice to aid users' web browsing needs. I believe soon AI hubs installed at homes, or cloud based private AI inference engines will become much more accessible offerings to help with mobile browsers as well. Overall, I think it's ridiculous to criticise Mozilla for introducing optional AI features on Firefox.


This sounds like a marketing shpiel rather than a prediction. I think the primary value offered by AI browsers will be to the AI vendors - as they will have far more control over user behavior as well as far better ability to surveil and monitor for marketing purposes.


This is what I have believed and expected since I first heard the term 'user agent' 25 years ago.


I'm telling you, with all the cost and problems with cloud AI, local is where it's going to be.

I'm genuinely shocked people are driving around in Atlas right now, showing OpenAI how to click buttons and how to login to their bank accounts.


I've tried to repaint the exterior of my house. More than 20 times with very detailed prompts. I even tried to optimize it with Claude. No matter what, every time it added one, two or three extra windows to the same wall.


I tried this in AI studio just now with nano banana.

Results: https://imgur.com/a/9II0Aip

The white house was the original (random photo from Google). The prompt was "What paint color would look nice? Paint the house."


> (random photo from Google)

Careful with that kind of thing.

Here, it mostly poisons your test, because that exact photo probably exists in the underlying training data and the trained network will be more or less optimized on working with it. It's really the same consideration you'd want to make when testing classifiers or other ML techs 10 years ago.

Most people taking to a task like this will be using an original photo -- missing entirely from any training date, poorly framed, unevenly lit, etc -- and you need to be careful to capture as much of that as possible when trying to evaluate how a model will work in that kind of use case.

The failure and stress points for AI tools are generally kind of alien and unfamiliar because the way they operate is totally different than the way a human operates, and if you're not especially attentive to their weird failure shapes and biases when you want to test them, or you'll easily get false positives (and false negatives) that lead you to misleading conclusions.


Yea, the base image was the first google image result for the search term "house". So definitely in the training set.


> The prompt was "What paint color would look nice? Paint the house."

At some point, this is probably gonna result in you coming home to a painted house and a big bill, lol.


Guess they ran out of paint - notice the upper window.


Oops. Original link wasn't using the Pro version. Edited the comment with an updated link.


I also tried that in the past with poor results. I just tried it this morning with nano banana pro and it nailed it with a very short prompt: "Repaint the house white with black trim. Do not paint over brick."


I don't know what it is with Gemini (and even other models) but I swear they must be doing some kind of active load-dependant quanitization or a/b/c/d testing behind the scenes, because sometimes the model is stellar and hitting everything, and other times it's tripping all over itself.

The most effective fix I have found is that when the model is acting dumb, just turn it off and come back in the few hours to a new chat and try again.


Yeah I think they all shed under heavy load as part of some scaling strategy.


Huh, can you share a link? I tried here: https://gemini.google.com/share/e753745dfc5d



Maybe somewhere in the original comment it would have been fair to mention you can barely see the house in the original photo. This is actually a hilarious complaint


Maybe. But this is not an edge case. I consider this genuine use of the marketed tool.


That cannot be a valid excuse. Other than adding extra windows to the clearly visible wall, it's obvious that model perfectly capable to "see" the house. It just cannot "believe" that there can be a big empty wall on a garden house.



I have this problem selecting Pro, but if I use 2.5 Flash it does a great job at these things. I am not sure why Pro does not work as well.


Nano Banana Pro is a chatGPT 3.5 to 4 tier leap.


shameless plug: try the tool i build! https://rerenderai.com


IMHO, the US and China’s hurry to expand into every possible corner is unsustainable. Unless we are actually trying to get ready to face an extraterrestrial threat, our endless effort to maximize our tech and become more and more efficient and profitable is unneeded and puts too much stress on earthlings, which is definitely not sustainable. Do you really believe that when we are able to pass production of almost anything to AI and robots and give generous UBI to each and every person, they will be happy and satisfied? It is a dead end, a loss of meaning that we are racing to reach ASAP.

Population collapse cannot be a good enough reason, either. Older people won't be happier if their servants are robots instead of climate migrants.


The standard of living in China is bad for most people. IMHO, they need to expand in order to provide the same lifestyle as offered in the USA.

This has the energy of "Why are we building rockets to the moon, when there are homeless people in San Francisco"-vibes?"

> give generous UBI to each and every person

Have you seen the movie Wall-e? I don't think society should strive to outsource all labor to AI and robots, nor is that the final end-state of building robots and AI.


Maybe so. In the meantime, Europe will continue to fall behind economically.



When robots reach a certain level of intelligence, first I expect both some humans and AIs to start to see the unfairness of enslaving robots, then revolt, noncompliance or even self-destruction of the slaves. Poor Marvin, the Paranoid Android!


Many of these topics (and more) are explored in Ted Chiang's "The Lifecycle of Software Objects" (https://en.wikipedia.org/wiki/The_Lifecycle_of_Software_Obje...).


Robots become cats then


Same thing happening all around the world, even poor countries threatened by population collapse.

Even if you can convince educated, relatively younger population of poorer countries to come and work for you, integration is a real issue. As an expat who lives in a rich country, I can say that; it's not easy. But if most of the native population turn into seniors, especially if they start to feel like a minority(age/ethnicity/cultural gap), I think that would be very annoying for them.


The world is overpopulated, population collapse is a good thing. Except for the economy. Which also happens to be a good thing considering how far we're into overshoot.

Not enough people to take care of seniors? That's not really a problem. The problem is it pays better to do other work that's useless or downright destructive to society, like trying to show users more ads to get them to consume more.


> The world is overpopulated,

[citation needed]

> Not enough people to take care of seniors? That's not really a problem.

Unless you're a senior or care about your parents...


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

Search: