highly recommend trying to build an app with maximal AI usage (ie AI does most-all of the coding)—it's pretty eye-opening.

you a) learn a lot about how to use LLMs effectively
and b) come to appreciate just how much good the models are getting

last night was when I crossed the threshold from "AI is okay at coding but it's not that great" to "wow this is better than most engineers, including me"

design @splits | /wireframe /latenightcrew /event-pass /feud

I don't think that necessarily means all jobs are gone in 5 years. 

But I worry specifically about new grad employment. Why hire an engineer straight out of school that you need to train, when you can have an AI that can do most of the job right now?

Design and creativity in general is going to be 10x more important. Good or great design is innately human, imo. LLMs are decent at getting a rough product, but horrible and creating a wire frame or a logo from what I’ve found. 

Also the BD guy that can get people to like him and his product might become more important.

Helping build @midao. Probably on the water if I’m not thinking about crypto. /eagles /betting /off-grid

For sure, I think it’s really good at giving you something rough and generic. Typically good enough to use right away if you’re not picky. 

But when I’ve asked an LLM to create a logo, there’s never anything unique, and usually more prompts make it worse. 

To your point, I’ve found LLMs to be best used for feedback and implementation than the ideation phase.  

Also I think a good copywriter with a unique voice will go far. AI slop has such a unique style of sucking the soul and authenticity out of a product.

Creativity yes, agency yes, taste yes. But I think a lot of what we think of as "product design", AI will be really good at that too, and soon.

FWIW I think it's actually pretty good at logos. It's not great at wire framing but it's getting better at being able to critique them and offer suggestions.