Working out what you want to do, a visitor from Wales & more…

Posted by

On 25th September, 17 freelancers met in the Battle of Trafalgar pub to talk all things freelancing and technology.

This week we were joined by Jeni for the first time in years, as she moved away from Brighton and couldn’t come back during the pandemic. It was great to see her!

  • A very welcome visitor from Wales
  • Losing your main client
  • What will LLMs do to work? Will LLMs top out soon?
  • Looking for projects without the help of Wired Sussex
  • In a parallel world, what are three jobs you would do?
  • Becoming a teacher
  • Looking after your mental health
  • Using LinkedIn to find work
  • Making indie games
  • Rich people and blood transfusions
  • Bitcoin prices over time
  • Don’t con yourself that you would have sold out at the peak of a speculative investment
  • Visiting Brighton as an ex-resident
  • DrupalCamp Cemaes
  • Putting on events
  • Keynote/Powerpoint Karaoke
  • Mastodon and federation
  • The changing world of social media
  • Tiktok is like crack
  • 3D printers
  • “You want me to buy one that is character building”
  • Macbooks – go for a Pro with default RAM or Air with more RAM?

Highlights

Working out what you want to do

Peter had a great tip for people struggling to work out what to do or how to place themselves when marketing: In a parallel world, what are three jobs you would do?

This was an exercise he’d picked up from somewhere that I failed to write down. The idea is it slightly divorces you from being tied to what you do now or what you have done, and lets you think about what you could do/have done. Useful for deciding what you want to do if you want to change, but aren’t sure where you should go next, whether that’s the type of work you do, or something else in life. I think it could be useful for people who do many things and want to narrow down how they describe themselves when marketing.

AI/LLMs

Large Language Models (what is often meant when companies talk about “AI” currently) are coming up in conversation a lot recently. We talked about their usefulness within writing code, and also what some of their wider impacts might be. This let me roll out a pet theory:

What if LLMs now are 80% as good as they’re going to get?

We’re seeing a lot of companies use LLMs, and we’re seeing a lot of companies put access to LLMs in their products, whether requested by their customers or not. Some people think they are going to develop into something that can be regarded as fully intelligent, i.e. more what is called “AI” in science fiction (“AGI” as industry now has to call it as “AI” is used for many things – Artificial General Intelligence.)

What if… Large Language Models are going to get better, but only by a bit?

For programmers, LLMs can already be useful, so a slightly better version means a slightly more useful tool. Terrific.

If you’re expecting to be able to replace staff with an AI tool, one that’s a bit better? That’s not going to be good enough to replace a person, unless the job is so simple that you’ve already replaced them another way, or that your understanding of their job is so poor that you think they can be replaced when in actuality, they can’t.

There are plenty of complaints out there of customers of large companies finding their only way of talking to the company is via a chat app on the company website, and then they are having strange interactions with what turns out to be a bot backed by an LLM, rather than a person. Will they be helped by the LLM being a bit better? Well… probably. At least, some more queries will be able to be handled by the company. If the company is big enough, that might mean they can get by with a few less staff, if it’s more affordable to replace them with an LLM tool.

If this is the way things go, it will be interesting to see how many LLM providers can survive. Will there be enough use cases to keep them going? Apparently OpenAI, creators of Chat-GPT and other LLM based tools, is currently losing over $5 Billion a year. It’s going to take a heck of a lot of customers for them to become profitable. If they raise their prices so they can become profitable, will the customers still think the tools are worth it compared to hiring a person, who will always been the more adaptable solution?