This piece from Bill Gates is a well thought out analysis on what is coming for us with AI. The cynic in me focuses far too much on the negative effects, and the negative effects are most of what you see in the media these days. I use AI every single day in my job. It is absolutely like getting to wear a super powered mech suit that allows me to do far more in far less time. The current frontier models still have very poor judgement though, and frequently ignore stored memories with explicit instructions from past conversations (so very human of them).
I hope we get to see the economic and technological advances Gates envisions. That would be a very nice future for my kids to grow up in. I also hope saner minds are elected to government so we can avoid the worst effects to come.
I think Cloudflare has a pretty amazing product with Workers. It has rekindled the same ease of development and deployment Rails had with Heroku a decade ago. There are several frameworks (Astro, Hono, etc) that make it easy to write small, simple applications in Typescript and run them on Cloudflare with a single command. It helps that their free tier has a very high ceiling.
There are some gaps in the platform, no geospatial functionality on D1 for example, and that’s okay. I can’t migrate everything to Cloudflare and be absolved for sysadmin responsibilities. I do hope that Cloudflare gives me a reason to really consider that possibility one day. I would thoroughly enjoy never having to manage servers and all the network topology that comes along with it ever again.
I’m fairly confident that I will rarely write code only by hand anymore. While the latest frontier models still lack good judgement when it comes to accomplishing a task, they are more detail oriented and persistent than I will ever be.
At the same time I’m not sure I’m comfortable letting an AI agent work completely autonomously with access to GitHub to create, review, and merge PRs.
I like to see what the agent is doing, and retaining push/deploy permissions for myself.
I’m sure even this safeguard will fall by the wayside as frontier models become more capable with decision making and judgement.