Design a PCB with DeepSeek

Hello everybody.

Here you can see a good tutorial for design a PCB with DeepSeek :

Best regards.

Jean-Pierre

2 Likes

Hi :smiley:
Not really my ā€˜Bag’ I must admit as I like the challenge with just the rather limited resources at my disposal and that’s squidgy and grey and resides between my ears I’m told :wink: Thanks for the post as this is something that will interest the younger players and it’s nice to keep things current and relevant :grin:
:mouse:

This is not a PCB design with AI tools.

It is exactly what you are referring tutorial information from an LLM, and then you design by hand.

Please be clear, do not confuse.

Maybe that one is baloney, but a few weeks ago an OP mentioned some experience with an AI based PCB layout program.

Off on one end of the opinion spectrum, there is a school of thought (interviewed in New York Times opinion) saying that all of our jobs (more or less) will be taken over by AI in about 2 years. I can forward a link to anyone (but not too many!) sending me a PM.

I am skeptical of this but I would not bet a lot of money against it. I do wonder why nobody has a robot to chase deer away from our garden, or dig weeds, or clean moss from our roof, and those seem a lot simpler than most of our jobs.

Maybe many of the jobs at the NYT can be replaced by AI in the next 2 years :wink:

2 Likes

I cannot imagine how an AI robot in my lab could find, retrieve, place, and solder parts onto a pcb.

This has been happening for a long time in warehouse accounting programs, and in large enterprises, there is also automation of storage. All this resembles the evolution of mechanical engineering. A voice assistant, not a replacement for an engineer.

Like fusion, that is ready in 20 years, AI will replace all jobs.

So why doesn’t AI solve fusion and design a working reactor?

Of course it won’t replace everything. But it will reduce it significantly. As now for 3 machines you need 1 operator instead of 3 turners))

How do you know it hasn’t? :wink:

Simply by evidence.
It is enough to count the number of fusion reactors that output more power than is being input and does it for days. You can ignore the AI-part until you reach the number one.
Wait 20 years, repeat.

You mean like the Stellaris that solved a geometric puzzle that humans couldn’t to realize a more optimal containment structure compared to the classic Tokamak. Theorised in 1951 but couldn’t be solved and it wasn’t until 2018 that it was getting possible and it produced plasma in 2023 with a reactor construction due to start next year.

Proxima Fusion has built on decades of research from the Wendelstein 7-X stellarator , using AI and 3D printing to improve the design beyond anything previously built.

Germany’s Nuclear Breakthrough – This Could Surpass China and Russia! – Engineerine

But this is going SOOO offtopic now I can’t even remember what was discussed

You mean Wendelstein 7-X. And humans did theorize the geometric puzzle …

[quote=ā€œNaib, post:12, topic:60739ā€]Theorised in 1951 but couldn’t be solved
[/quote]
due to lack of computational resources available. No AI involved. 1951 was the time of slide rulers and IBM big iron with hand woven core memory. The geometry was computed before 1975.
Wendelstein 7-X does not produce constant power and is not AI generated. But you already showed that by yourself.
Wendelstein is a demonstrator and ITER is a demonstrator. The first Wendelstein stellator was built 1975. Design of ITER I was finished 1998, first power is planned for 2034.
Wait 20 years, repeat.

I already deleted posts from another thread which had speculations and subjective opinions about modern AI. Let’s keep the focus in KiCad.

Were we discussing something?