Normies aren’t going to download VS Code and start creating production ready apps. It will be platforms like Bolt, Lovable, V0 or something like that which will allow people to increasingly create more sophisticated / usable software

I think you jump to conclusions prematurely. While I agree with your take I also think people will create new interfaces that allow people without the implicit knowledge and understanding of programming to operate these agents in natural language.

This is great for developers however, I've read that the model has regressed in most non-coding related benchmarks and vibe-checks. Is there a reason Google won't just do a specialized coding model and a general-purpose model that is better for other use cases? Maybe in the future?

I won’t respond anymore since this seems like low effort reactionary takes. I don’t even think you looked at what I’ve shared.

Have a nice life.

Social Media directly led to the rise of Fascism in the US (and possibly worldwide) are you going to delete your account now?

Do you actually know how real artists deal with something like this? By engaging with it. Maybe read about McNeal by Ayad Akhtar.

You think this is theft youtu.be/1ubjbJfRKV0?...

Some 15 year old kid in 10 years is going to make something beautiful that would’ve taken 100s of people without AI

You can politicise everything but the truth is OAI didn’t create this. People were and can do this via LoRa fine tuning for over a year now. They made it accessible to the average person.

This has nothing to do with Facism. Billionaire can and should be blamed for many things but this isn’t it.

After having used this model for a day I think its fair to say this is the best coding model out right now. Not just the quality but the speed too; is phenomenal.

I hope it is priced reasonably and has better rate limits and availability then Sonnet 3.7 (also a great mode).

I think it brings up important things that we have to think about. Can you really copyright a specific art style or a particular representational artifact (that is not copied 1-1). I'm not for it, I don't think there is any inherent value is derivative works apart from their novelty value.

I can copy and paste.

I think there are multiple things going on here. I doubt anyone is making money selling or is using these in a non-ironical way to pass it as real art. I think what constitutes art is not as straightforward, change is coming. I've been thinking about Kafka's Josephine the Singer.

Its just a silly meme. Photography didn’t make painting obsolete but it did influence its future direction.

Have you seen this : youtu.be/1ubjbJfRKV0?...

Thank you. I've been really enjoying working with Gemini 2.0 Flash with the new Canvas feature and this is going to be even better.

I hope the overall agentic capabilities are also improved (function calling etc)

Is this only for those that have signed up for Duet AI or is it free for all google workspace customers?

I'm using openrouter (an LLM api provider) to access and use multiple LLMs for the translation for a personal project I'm working on.

Will be adding some features based on the insight from this post.

How is this legal?

Isnt it simply an implementation problem? I’d assume a screen sharing and webcam enabled multi modal mode would be able to get around cheating in remote tests etc. I’m working on some prototypes on this for an AI Native LMS so I do think there are positives to it. We got to find the right UX.