I went to an AI tech conference and it wasn't all hype

I went to AI for the rest of us in London this year - mostly because a very good friend invited me to his talk.

It wasn't just a tech talk and if you have a moment to watch it once the talks are up online I highly recommend it. That said - it does contain themes of domestic abuse and terminal illness so proceed with that in mind.

What was the vibe?

Overall I found that people were a healthy mixture of skeptical, curious, and realistic. There was also a broad mix of people within and outside of tech roles - it really did feel like "for the rest of us" was reality.

I mostly stuck to the tech track outside of the keynote speeches as that felt the most directly relevant to me in terms of work, but probably loop back to watch more of the talks once they're online.

Were all the talks I went to good?

Yes, actually - some better than others but in particular Charles Humble's talk on Green AI and Rory Malcolm's stood out to me as really insighful talks on topics I floundered to get a grasp of before hearing them.

Would I go to another one if I didn't get invited by a friend?

Yeah, I think so. I felt like I was learning rather than being reassured by people more optimistic than me.

What things did I take away from it?

Somehow, no stickers. There were many there but I didn't reach for them.

One neat trick

If you're going to code with an AI agent then configure its instructions to make it ask you clarifying questions on what you've said you want until you explicitly tell it to build the thing with the level of detail you've given.

This actually really works well and seems pretty transformative for that style of working.

I also really like the way it puts the human back in control - you're doing the telling and it's doing the asking.

A pile of links and resources I pulled up on my phone

• https://dora.dev/ - Project led by Google, somehow not an acronym
• https://shows.acast.com/visionaries-rebels-and-machines - this Jamie Dobson guy gave an amazing talk on the history of computing particular to machine learning and I intend to listen to his book and/or podcast
• https://www.braintrust.dev/ - great resource for looking at prompt flows and evals (i.e. how can we possibly get real work done with non-deterministic systems?)
• https://www.langflow.org/newsletter - low code building with AI, these folks did some interesting demos at the conference

Feelings

I feel slightly less like the mad one in the room for being skeptical while trying not to be cynical.

A lot of very smart people who've done this tech thing a lot longer than I have really don't know what's next but have solid ideas of what they want to see happen and how to influence that.

Most agree that we're inside something like the dotcom bubble, and one speaker put it very well - it's never going to be cheaper to learn these skills than it is right now, so make the most of it.