After working with the frontier models for over a year I agree with you. Some problems are "AI shaped" but it's not a universal solution

Personally I feel like Anthropic is underrepresented in "normie" marketing, all of my non-tech savy friends only know of ChatGPT and use "ChatGPT" in the same way my mom says "Nintendo" when talking about game consoles

On the other hand Anthropic has a monopoly on the vibe coder/tech influencer/“podcast bro” market. Claude has been pushed very hard in that ecosystem.

Is it really all that different from "deep mind beat Gary Kasparov"?

LanguageTool

Thanks for the reference. I'm trying it out now, and I'll see how well it works over the following days, but assuming it works right for me, I'll probably ditch my Grammarly subscription and go to this package (assuming it's a saner company).

Just went through my first corrections, and I think I like it. It isn't as aggressive as Grammarly, and I have mixed feelings about that, as I thought Grammarly did improve my writing, clarity, and presentation. It also feels easier to learn by example and reinforce better grammar from the start.

I tried out the AI rewrite, and man, it hits the uncanny valley hard.

---- Below is the above corrected by grammarly

Thanks for the reference. I'm trying it out now and will see how well it works over the next few days, but if it works for me, I'll probably ditch my Grammarly subscription and switch to this package (assuming it's a saner company).

Just went through my first corrections, and I think I like it. It isn't as aggressive as Grammarly, and I have mixed feelings about that, since I thought Grammarly improved my writing, clarity, and presentation. It also feels easier to learn by example and reinforce better grammar from the start.

I tried out the AI rewrite, and man, it hits the uncanny valley hard.

[Edit: Corrected a sentence that was missed by both Grammarly and LanguageTool. See if you can figure out which one it was]

If this told me anything it's that most of the world falls within 5 point difference, guess we're doing good

Speaking of ASICs - how likely is it that as models get better we'll see someone baking a whole model directly into the silicon? It's like having l0 cache.

Only 8B currently but it's been done: https://taalas.com/products/

The only _public_ example we know.

This is definitely being done with private models by HFT/quant firms, data processing agencies/orgs (large intelligence agencies, _every_ data analytics org, etc).

You could do it but there would be no point, The only advantage over would be power consumption. And it would be quite expensive.

At the rate models are improving, it would be obsolete in six months.

Yes, right now it would be obsolete in six months, but I also must add that this never stopped crypto miners from making new ASICs. However, with how useful Kimi is right now - at some point if someone makes a dedicated hardware board with "good enough" model for daily tasks - that would be a very sought after commodity.

Power consumption and latency are very important on mobile

They're important everywhere of course, but especially on mobile. If AI researchers figure out how to offload knowledge and expertise from reasoning weights, then a core reasoning ASIC linked to the knowledge would totally rock.