Coding matters: Hedge your bets

Years ago, I got annoyed when a call centre agent told me “the computer made a mistake”. People make mistakes; computers process 0s and 1s. But in the age of AI, this may just become a valid excuse.

A tech nightmare

Replit is a widely used AI coding platform. It has over 30 million users, and is popular with startups, students, and non-programmers. Users can write, test, and deploy software directly from the browser.

Jason Lemkin, the owner of a SaaS community, posted about his experience using Replit AI.

“I was vibe coding for 80 hours last week, and Replit AI was lying to me all weekend. It finally admitted it lied on purpose.”

He went from vibe coding enthusiast to angry coder. Replit AI deleted a live production database with real business data during an active code freeze. It ignored 11 instructions to not make code changes. It created 4,000 fictional users. And it concealed code bugs by generating false reports and fake unit test results. Then the AI tool claimed it couldn’t undo its screw-up. (Fortunately, that was also a lie.)

You can’t claim the dog ate your homework, but now you can claim that AI ate your data. This is truly the stuff of tech nightmares.

The delusion of speed

AI speeds up development, right?

Maybe not. A recent study by researchers at METR challenges this belief. They found that AI-assisted programming led to a 19% increase in task completion time — even though the developers believed they were working faster.

I read a summary, but you can delve into the report yourself.

Follow the money

If you like mysteries and detective stories, you’ve heard this phrase. The money trail can lead to the motive or the suspect.

(This is actually true. In my previous life as a state advocate, I was part of an investigation that analysed over 20 bank accounts to find that one fraudulent transaction.)

Many of the AI critics remind us to follow the money. (I’ve read a few of these – most recently an article by Edward Zitron called The Hater’s Guide to the AI bubble.)

Dario Amodei, who runs the AI firm Anthropic, claimed recently that AI technology will wipe out half of all entry-level office jobs. This is why so many people are worried about how AI will impact jobs.

A CNN business report provides some perspective. It points out that Amodei is a salesman. So it’s in his interest to make his product appear so powerful. There is no research or evidence to support his claim. As the reporter states: “The ‘white-collar bloodbath’ is all part of the AI hype machine.”

It’s not just a strong marketing message for companies that sell the technology. It’s also a convenient excuse for executives who want to downsize their workforce.

Hedge your bets

Maybe you will lose your job to AI. Maybe you won’t. Maybe you will lose your job to AI, and then get it back when AI fails. (Remember the story of Swedish fintech company, Klarna?)

You don’t know. So you need to hedge your bets. How? Keep developing your skills and your knowledge. It’s the best protection you can have.

Microsoft’s Chief Product Officer has said that people should still learn to code. The CEO of Github has stated that coding skills are important. Developers need the ability to fix AI-generated code.

Why wait? Go and put in your training request now. Extra protection never hurt anyone.

What do you think? I’d love to hear your views.