[zloy88]

Maybe it's a way of perspective. I adore to use AI to actually learn, so I don't feel like I offload thinking. I use AI to do all the research which earlier I did manual through google search. I still make my own decisions, I just let Claude spoon feed me all infos I want and need. Feels great man

[hasteg]

Using AI to actually learn is indeed very helpful I've found... I just don't think most people use it for that. I recently wanted to build and train my own (albeit, small and simple 10M param model) and used AI extensively to explain concepts for me, explain lines of code, and generate in depth visuals I can use to get a visual intuitive understanding of what is happening under the hood. I think people who have a natural curiosity to understand the why and how of things benefit immensely from it, but I do admit it is very easy to just offload all thinking to it instead of asking it why something is happening. I could have just asked it to implement the entire model using PyTorch and just ran the CLI command to begin training on some auto-generated dataset, but intentionally struggled through it. Actually learning from it requires intentionality for sure.

[Jyaif]

If you build stuff with AI it's different. It's very tempting to defer many (too many?) decisions to AI.

[zloy88]

I do daily. So far I am doing fine, not sure what exactly your point is, sorry.

[user43928]

Same here.

If anything, I think most here outsource too little thinking to AI.

What am I supposed to be afraid of? Losing skills I no longer need to get the job done?

[zloy88]

Yea, kinda agree. I feel like my job just changed into a solution architect and quality assurance inspector.

[timmytokyo]

"Solution architect" sounds like a fancy term for a glorified button-pusher. At this point you're making yourself interchangeable with a bunch of other low-skill button-pushers.

[timmytokyo]

Why should anyone ever hire you? We'll just hire the LLM you use.

[user43928]

By the time the LLM operates autonomously without my input, I doubt your chances for employment are going to be any better than mine.

[Jyaif]

For one, as of today SOTA LLMs sometimes take the wrong decisions

[zloy88]

For me thinking and making decisions are two different things. I rarely let AI decide, unless it's something i do in my private time. But at work, I am the expert and I maximum let AI give me options, but I decide.

[brazzy]

This is also now the first thing that I find it truly useful for in a non-coding role: researching how to do things in Azure, which I have not used before.

[warshinder]

It’s just so sycophantic though.

[zloy88]

Lol, wdym. I like to prompt it that I want evidence, so when there is any new topic where I currently don't know how to tackle it myself, I ask that it does a research and give me evidence. So i can check the sources myself. Like I said, I let it spoon feed me.

When I know upfront how to do anything, I just give all the instructions. But the OPs point was If we offload thinking too much, so that's why I was just thinking about this example when I need thinking - that's usually when I need to learn something new.

[al_borland]

I've found it present wrong, skewed, biased, or old information. In other cases, it seems to ignore the elephant in the room. I find all of these things problematic when trying to learn something new. The less I know about a subject, the more susceptible I am to being mislead.

When I bring these things up, it will apologize, tell me I'm right, and adjust. But what if I didn't know enough to question it? Approaching from the other angle, maybe I'm actually wrong and it's a sycophant, as mentioned, trying to please the user?

On the topics where I'm having to correct it, I probably shouldn't bother asking in the first place. On topics where I'm not correcting it, is it because the AI was right, or I just don't know enough to call it out? This kind of thing worries me about AI being leaned on more and more as a teacher.

[zloy88]

Never rely on the trained data. I always ask to do a research which means, it pulls fresh data from the web. Another thing I enforce all the time is simply "don't make assumptions, give evidence" - that basically means it has to prove from where it takes it's statements. I do pretty fine like that.

[scotty79]

I don't understand people who can't get past that. I just do the same what I do when humans compliment me. Gloss over sugary bs and focus on the content.