Met with Dan yesterday. It was a great conversation.

First, I forgot to tell him we were expecting :) which is related to my thoughts on AI.

I think there are a ton of opportunities coming from this AI boom: securing AI, more efficient communication protocols and intermediate representations, etc. It’s a tool that enables senior folks to move at incredible speeds. To quickly, asynchronously and agenticly, kick off some work in the background. This ideas that used to sit in your head that you wouldn’t do because you didn’t have time. Now there’s no excuse. But is this a good thing? It’s kind of like DFW’s Infinite Jest and perfect entertainment and this notion of perfect satisfaction. Is this a good thing?

AI undoubtably has been an amazing tool. If you fully buy in and give it access to everything — which I still wouldn’t because most of the security and access and data sharing is still unknown — you can see tremendous efficiency gains. Imagine: someone opens a Jira ticket for an issue or feature request, this kicks off a workflow and AI opens a PR with the solution and you (as a human) review it. This is where we are. I don’t think we are at a state where we can reliably remove humans as reviewers yet. And this introduces a new bottleneck. Now the bottleneck isn’t in writing the code, but reviewing it. We are feeling the pains and fears of many years’ lack of investment in testing and verification. Should we as engineers be fully focused on doing QA now? Building out verification processes that we fully trust.

I’ve already offloaded all of the things that I don’t want to do to AI. Doing things that I’ve already done a few times and don’t feel that I have much to gain from. I’m still holding on to learning where I can — but there’s a weird dichotomy. Do my superiors care to invest in me learning the fundamentals or should I be learning how to do this with AI?

We’re also seeing a shift to seniority again. It used to be you wiped out the seniority to replace with cheaper laborers. Now, the experience is valued because you need to guide the AI and guiding AI is easier than guiding folks with less experience.

I took a lunch with a local manufacturer. They are still doing “old school” software engineering. It’s a three-person engineering team. Is this the future if you want to feel invested in and have a sustainable, long-term career? Most “big tech” is less interesting to me now. Everyone is trying to satisfy the market, move faster, make more; rinse and repeat. A “rat race”. A “race to the bottom”. The disinterest in small companies was ego- and pay driven. But I think my perspective on both ego and pay has shifted. Or at least beginning to shift.

However, with all of this said, I’m not sure I’d accept this tool given the potential associated costs and disruptions. Like code that I’d reject, I think it changes too much all at once. Too many unknowns. Too few ways to protect against what could be “one way doors” or irreversible changes. We needed a better Google. I don’t think we need agentic. I think we should be working toward an AI that guides humans instead of AI that does the doing.

I feel lazier. It encourages context switching and discourages deep focus. I don’t fundamentally believe this is a good long-term shift for thinking. For doing, maybe; but definitely not thinking.

We’re already seeing scary things happen or be asked of these super powerful tools like this where the government revokes contracts with Anthropic because Anthropic feels responsible to speak up and not allow surveillance of US citizens or AI to make fully autonomous decisions controlling weapons. Amodei identifies that our policies are significantly behind the pace of the technology — which is scary; it’s the same thing we saw with social media a couple years ago.

Here is another good state of concerns video by Hank Green. I haven’t finished it yet, but the first 30min hit the nail on the head.

I think I’m still hopeful. But at the same time, I feel pulled away from investing in my career as much. I feel the need to pour my energy in family to calm the anxieties of today. Is this selfish? Is it selfish bringing a kid into this world?