Comment by al_borland

14 days ago

> how do you usually work on tasks that you find boring?

Often times, I don't... until it becomes a problem.

If something is the right balance of boring and tedious, I actually like it, if I get started. It becomes something to keep my hands busy while I zone out.

> Which AI has been the most useful to you in your opinion?

I mostly use the Kagi Assistant, which defaults to a Kimi model. I occasionally switch the model (it offers many), just to see if it's better. I'll sometimes use ChatGPT as well, if I am doing something where going beyond a chat is helpful. So far they all seem to have similar weaknesses. Yesterday I was asking it something about baking and it said some stuff that sounded kind of dumb. I asked for more detail and then in contradicted itself in multiple places. When it does stuff like that, which happens a lot, I can find it rather paralyzing if it can't clear things up. I want there to be a "right" answer. When there is a debate on what is right, and it phases it as two different absolutes to create the contradiction, that's a problem. I'd rather it be up front and phrase it has a debate, give the reasons why each side takes their position, the trade offs, and then a recommendation based on circumstances. On the other hand, it will give a wishy-washy answer based on an obscure technically, creating debate where there really is none. It's a hard line to walk and I find that to be the biggest stumbling block I currently have with AI (outside of coding, which has many more stumbling blocks).

I see... that totally makes sense! I sometimes love that too, especially when gyming when I rely mostly on my muscle memory. I use that time to listen to podcasts or videos I usually dread watching.

interesting! I've never heard of that model before, so I'll check it out!thanks for sharing. I guess your issue with AI is mainly the accuracy then I suppose.

  • > I guess your issue with AI is mainly the accuracy then I suppose.

    Yes. If I didn't care about accuracy, I'd just make my own assumptions and hope for the best. If AI can't be accurate, what is the value?