A Brave New World (Of AI)

today we launched ChatGPT. try talking with it here: https://t.co/uWra8LKFMN

For modern AI there was Sam Altman. Todays LLMs have gotten good. We still don’t really quite grasp how their emergent capabilities work. Large parts of daily developer life have been erased in the last year and will most likely never come back.

There was a time where you could burn the midnight oil tracing a bug or reading to understand a system and this form of deep thinking was valuable. Today companies expect engineers to ship more code faster and leave the coding to the LLMs. Some are even asking engineers to leave the review to the bots as well.

There is a certain liberation to knowing that we’ll most likely never need to use the human brain to reverse a binary tree ever again. But where does this leave software engineers and the field as a whole? There is a tremendous amount of uncertainty at the moment for many good reasons.

Computer science and programming up until this point has been functionally deterministic. The user experiences and operating systems we’ve built up over the decades are based on that expectation. LLMs change that deterministic behavior and open new ways to interact with technology. I don’t think we can take for granted that layering this new technology on the old is the right sequence and there’s a lot of development happening to explore an Agent OS.

Software is dramatically changing and with that our role as engineers. There’s a lot of ways to think about it but for me the exciting part about programming was never about typing, it was watching ideas come to life and feeling like the computer was a magical canvas. LLMs make that imagination to canvas pipeline dramatically faster.

Engineering was always about finding a problem and inventing a solution using a tool (e.g. a computer). That part hasn’t changed, people still have problems that need solutions and a computer is still a multipurpose tool. But without a doubt, what has changed is the SDLC of an engineer.

Unfortunately many treat AI as a magic black box function where input goes in and the answer comes out. We learned with the rise of social media that software can have ripple effects across society. We will certainly be learning new lessons with the increased offloading of cognitive labor to machines.

We’re also learning that much of our technology up until today was a house of cards. 70% of the internet was exposed overnight due to a vulnerability in PHP. We’re certainly living in interesting times.