University, AI help laziness and hardcore tech, notes from 29th Jan

Posted by

On the 29th January, we seven freelancers got together in the Lord Nelson pub to talk all things tech and self employment. It was a bijou meet up, but the chat was still good.

Some of what we talked about:

  • Circular graphs in PostgreSQL
  • Site scraping & RSS & APIs
  • Is there too much box ticking teaching in university courses now?
  • Teaching your child more than their university course is
  • The delight of being at university
  • When writing your own data store makes sense
  • “I’m worried Chat-GPT is making me lazy”
  • SSL certificate supply and provision
  • SSL certificate length reducing to 90 days as a step towards 7 days and the problems this may cause
  • Writing your own programming language
  • TempleOS and the creators mental illness
  • Deepseek
  • Censorship in Deepseek and Gemini
  • Assassinations and Dread Pirate Roberts
  • Chip smuggling

Are AI tools making us lazy?

There are various “AI” tools based on Large Language Models (LLMs) which help programmers write code. Some people in the Farm love them as they feel they gain a useful productivity boost from them, some people hate them as the companies creating them have copied in lots of copyright material without permission and used a lot of electricity (and produced a lot of carbon) for what they see as marginal gains given the environmental impact.

Now, we seem to have hit a new phase: supporters rejecting the tools?

“I’m worried Chat-GPT is making me lazy” was the quote from someone I’ll leave nameless as I don’t believe they are lazy, and that’s the sort of thing potential clients don’t want to hear someone admitting.

The problem the person is finding is they’ve become over-reliant on prompting Chat-GPT to writing a version of code for them. They then re-write that code to be correct and fit their style, but are finding they just don’t write anything without the tool now. It’s the first thing they turn to when taking on a task.

This will be great news to companies like OpenAI, creators of Chat-GPT, who are currently losing money even with their higher paid accounts. The more people reliant on their tools, the higher chance they can ever be sustainable. But my friend is worried they’re fundamentally breaking their own way of working by over-using LLM tools.

Six months ago, Sul stopped using Chat-GPT while he is learning React as he found he wasn’t retaining enough of how to use the Javascript framework while using Chat-GPT to help in his coding. My friend is finding a different version of this problem. They’re a very good, very experienced programmer, but are solving problems in a way they can phrase to a tool, which is not always the way they feel they should be solving it.

Sometimes, you have to wean yourself off a tool, or go cold turkey without it, depending on your personality. This might be an aging but beloved editor that isn’t supported anymore, or a bang up to date tool you’re not sure is really working the way you want. You might come back to it, you might be happier without. It looks like the AI tools will need mass adoption by the sector to be sustainable businesses, so they’d better hope my friend does keep using them once they’ve had a break.