Thanks for making me feel older now

It cost about £190 in 2006.

Now we have GPUs that are in tens of thousands of pounds with insane performance, but what would their price be without the AI and Datacentre squeeze?

They are a consultancy in Germany, but I watched a presentation on them tuning and removing bias from Deepseek models. It was quite interesting.

https://www.tngtech.com/en/about-us/news/release-of-deepseek...

(I upvoted your question as I agree)

Its not just code we need to worry about, its also subliminal messaging and other things.

Do you have your agent say things like "Hasta la vista baby", or "I'll be back, after I clear my context" ?

    "permissions": {
    "allow": [
      "Bash(npm run lint)",
      "Bash(npm run test *)",
      "Read(~/.zshrc)"
    ],
    "deny": [
      "Bash(curl *)",
      "Read(./.env)",
      "Read(./.env.*)",
      "Read(./secrets/**)"
    ]
  },

Is the geographical data more useful, or are buildings and other structures more important?

Genuinely don't know much in this space.

I found LLMs good for spitting out a structure for my posts but if I ask it to write for me I end up just fixing or correcting it and then lose motivation and never finish the posts.

After a bit of soul searching I have come to the conclusion and reassertion that I want my blog to be about random stuff I like or do, not just a content piece of things I can say that I am a "thought leader" or "I live, eat and breath tech" during an interview.

Maybe in years to come people will just be better evolved to detect AI or slop and by then writers will have ways to show legit writing and captivate readers.

1. Chat

2. Autocomplete

3. Embedding knowledge using RAG

4. Tool calling by LLMs (CLI or MCP)

5. Agentic LLMs executing task(s)

What do you see the next step or iteration?

My theory is that we will get more quantization and efficient models by the end of 2026 and my hope is that we will have mini models that wrap around tools (I call them domain agents) that just give answers without bloating context.

i.e. the Domain agent gives the calling agent the sausage but doesn't explain how the sausage was made.

Curious what your theories are, but I think we might need a whole rethink of the architecture of LLMs being combined with tools etc.

Comments URL: https://news.ycombinator.com/item?id=48478162

Points: 3

# Comments: 2

It might just be the liberal bubble of London, but the sentiment I saw with some colleagues was.

- "Why would I want to serve, I am not a robot"

- "You have to have a screw loose to be able to kill someone"

This was in more eCommerce and design industry stuff in London, the defence sector could be more supportive of exmilitary, but I don't know.

Comments URL: https://news.ycombinator.com/item?id=48473104

Points: 3

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=48416959

Points: 3

# Comments: 1

Would you recommend this for hacking around or not?

We aren't enslaved or at any risk just yet with this form of "thinking machine", but we do need to culutrally and emotionally come to terms with them.

Let's not have any other "jihad" just yet, there's plenty of other jihads going on right now.

I just hope he doesn't start mailing packages to people in the tech industry in the next few years.

I was meaning more removing your reliance on Google in your life for search, photos, entertainment.

You can start small with Duckduckgo etc.