galaxyLogic

Good point. I've though about my own mental attitude when interacting with AI agents. When they do something good I feel like saying "Thank You". Does that make sense? I guess it does because it communicates to the model their output was correct. But it feels silly to say "thank You" to a machine. I guess I just have to get over it?

Isn't this much the same as OpenClaw or Hermes?

I think it's a good thing that AI providers are "coalescing" on an agent-model, by producing competing agent-products.

But so what would be the benefit of "Dots" over OpenClaw, Hermes, and Muse?

The problem I encounter with async/await (in JS) is that while an async method can call a non-async-function and do somewthing with the result of that, the reverse is not true, a sync function can call async-function but can not us the result of that in any way, except pass it on or upwards.

What makes it worse is that you can not simply modify a sync-function to become an async-function, if it has existing callers because those would likely break them.

This affects the whole tree of possible function calls in my program. If at some level I have a sync function but I see it needs to get to some data that only async fuction can provide, I may need to change a whole call-chain of my call-tree, not just modify a single function in that tree.

Then as I develop my program I need to make the decision for every function; should it be sync or async? In many cases it could be either one so which should I choose? Making it async would seem to make it easier to evolve the program later. But then would it make sense to make every function async?

The British rock-band Traffic said it in their "Low Spark of High-Heeled Boys" : "... we were children once, playing with toys".

I wonder if it's true that people don't remember. People maybe remember but don't want to talk about it. It is such a cognitive dissonance.

LLMs are "readonly" I guess for several reasons:

1. Technical cost of updating the mode.

2. Inability to trust every user's "truth".

3. Ability of AGENT-HARNESSES to learn with the help of the human user.

So agents learn, LLM already knows everything it will ever know, and ESPECIALLY it has already learned how to understand human language.

No 3. above means there is no danger of the LLM getting corrupted. But the agents running on user's machine learn on behalf of that user who shares the machine with them.

Can you not just tell them: Imagine you are chatting with a very knowledgable person who knows answers to most questions, and can do many computer-related things for you ?

“The solution is something we talked about for many years,” he wrote. “Do not trust the model output.”

Is it possible to know when a request comes from an end-user vs. a bot?

I don't think it's "funny" . But AI has brought it up to the discussion. In human organizations the decision makers make it so it benefits them. But with AI it is possible to design the whole delegation-structure by anybody who designs agent-collaboration.

So its becoming more obvious that AI can actually help big corporations organize their command-structure with the help of AI to produce more profits. The boards should take notice. Does it really help produce more value if most of it goes to the CEO?

Doing lots of programming over the years I've realized that I often need to write "code for myself", in my head, to think what I should do and in what order, and also why. I write to-do notes and notes on why something was best done or had to be done in a given way. Code is instructions for the computer, but developers need instructions too.

It always seemed to require extra effort to "program myself", maybe because it requires questioning whether I'm doing it the right way and what would be some alternative ways of doing it. It's almost like "out-of-box thinking".

Whereas basic coding is often as easy as writing this note here, just write what comes to my mind. Coding tasks are often trivial, but they must be done. But they don't really burden our mind too much, not too often.

But with AI, it's all about "programming the programmer".

And that requires more thinking, asking more questions like is this really what we need to accomplish, or would some alternative way be better? What alternative way?

Asking questions like that was always part of the work but with AI it seems to be the only type of work. And it is more difficult, more exhausting, than basic coding.

Meant to support 50 developers. But I wonder, would it make sense to have some or most of those "developers" replaced by AI agents?

Depends on how we understand the word "insult". Can you really insult inanimate objects? :-)

Figuring out how to get what you want from people would seem to me to be a very different skill than figuring out how to get what you want from AI agents.

For instance, how do you motivate people to work long hours, put in extra effort, feel proud of their work? How would you do that with AI?

Reminds me of current day politics. Lots of public statements which are obviously false, and you would think the politician knows they are false, but utter them anyway because they also know lot of their supporters buy what they are saying anyway.

Now politicians also know something about their supporters so they will adapt their statements to what they think they can get away with it. But, I wonder if this leads to a two-party-system where one party attracts stupid followers and another attracts the smarter ones?

In terms of AI, we might see LLMs specialized to attract more stupid audience and others meant to attract those who appreciate correctness and facts.

Interesting. So if the LLM is having a discussion with itself, am I paying for the tokens it uses for that?

Right, but could that also be because ... the more long-winded they are, the more you pay for their output.

An observation: You can never insult an LLM, but it can certainly insult you.

You can not insult it because it does not care, because it does not have "feelings". But you do.

But if you are managing AI agents you dont' need "soft skills" do you? You don't need to be especially nice to the AI, or symphatize with it, or have fun with it to build trust,

I would imagine that managing a team of AI-agents is totally different from managing a team of people.

Developing on the back-end has much more mature tools in my experience, IDEs and QA tools. Client-development is often not much fun dealing with the browser dev-tools. So I share your feeling thath serious development is better done on the back-end.

But in principle there should be a client-dev-tool as good as those for the back-end.

There is some progress. I've used WebStorm to do client-side work which has a niocve feature that as I debug my client-side program I can use WebStorm instead of browseer-tools and if I see a typo in my code while debugging it I can edit it away and it gets saved into the source-file where it came from.

But, if you can do things on the client (with JavaScript) you don't need to send (so many) requests to the server, making latency less of an issue.

Doing things on the client means user's CPU is doing some work which else would need to be done on the server, for maybe thousands of clients at the same time.

So I understand some people don't like JavaScript, but then I think the solution would be WebAssembly. I mean the point of distributed computting is that the computational load can be distributed. Perhaps counter-intuitively that often also means less need fo communications and latencies.

> translating "doBigDNNThingNVidiaGaveMeInAKernel()" to "doBigDNNThingByHandBecauseAMDDoesntSupportIt()" isn't a rote translation at all.

I wonder if this points to a deeper limitation of AI, it can not do coding tasks it has not seen in its training material. Or could it possibly "generalize" to accompllish something like this anyway?

Saw this on the web:

"AMD and Anthropic also formed a multiyear engineering partnership to optimize ROCm using Claude"

FROM: https://finance.yahoo.com/markets/stocks/articles/ex-amd-exe...

I rhink there's a difference with AI because -- it brings true value because you pay for the tokens, you only pay for what you use. That is true value.

Compare to just paying for an internet connection, you have bandwidth but not sure what you can do with it that is valuable.

Let's say you use AI to produce software. There;s no limit as to how high the quality you want your software to have. And how fast you want your project to be complete. There's plenty of room for higher quality, and more performant AI. As AI becomes chepaer people will use more of it, they're not going to say "We have enough AI".

Compare to railroads. Yes you pay for the distance travelled but there's a limit to how much people wwill want to travel, how it will benefit them.

Anthropic rewrote Bun in Rust, with much help from AI of course

And couldn't we just ask AI to translate our program in one language/framework into another?

I was under the impression that AI was supposed to remove software-moats, let us all ask it to write our custom MS Word for us for instance?

What I don't quite get is why can't they use AI to translate CUDA programs into more open architectures like AMD ROCm?

AI is supposed have solved the "coding problem". But shouldn't translating a program from one platform to another be an even easier, more mechanical, task for the AI?

Well it's easy to come up with code, not hard at all. But what kind of code? Does it work? And more importantly, can it be maintained and adjusted to evolving requirements?

Coding is easy just like writing is easy. What makes the difference is what you write.

"... the chip serve Meta’s Llama 3.1 8B at a blistering 16,960 tokens a second — when announced last February, that was 48x faster than Nvidia's GPUs and 8.5x faster than Cerebras' accelerators. "

I think the big news is that AMD is getting into memory-business so they won't be so dependent on Hynix and what have you. Memory is the bottleneck currently.

Feom https://brennan.day/what-have-note-taking-pkms-accomplished-...:

"German sociologist Niklas Luhmann, as he was famous for his use of the "slip box" or Zettelkasten note-taking method and the achieved output of more than 70 books and nearly 400 scholarly articles. This is the kind of remarkable output that many in the PKM community aspire towards."

I think that is somehwhat misguided. The goal shouldn't be great quantity of output, but great quality.

I think AI has revealed that to us in reccent years. AI can produce endless output. But who needs it, what is it good for? Unless we have a GOOD QUESTION to AI, it's output is not very valuable.

These note-taking apps, you write a lot of notes, but is there an overarching question the notes are meant to answer? Depends on how you use them of course. But I would guess that most of the notes are more "Good to Haves" than "Step towards a definite goal".

Most notes don't probabaly take you closer to the end-goal of your project. You just think they might become useful in the future, so you write them down. Just in case. To feel you are taking good care of your precious ideas. But how much time are you eilling to spend to read them in the future?