The Screen Is About to Get Smarter Than YouPosted by John on Feb 13th

I was networking at BNI this morning, and one of our lovely members made a throwaway comment at the start of the meeting during the open networking. She had read an article that discussed how AI was “coming to get us,” effectively, and she told us with a slightly nervous laugh. It was quite funny.

But it stayed with me.  Because I often forget how not aware of things a lot of the general population are currently.

So, I live in two worlds. I live in the world of business, where I’m meeting people and working with them all the time, whether it be on web apps or AI automations. But I also live in the learning world, where I’m spending a lot of time online, keeping up with the changes, working out where I need to have my focus and what I should be learning next.  So, I suppose I am aware of where AI is potentially going much more than many people are.

So, when I hear people talking about little about how they heard something about how AI might be dangerous, it interests me. Not because they are wrong, but because I am surprised they didn’t already know. 

But why would they?  It is rare that the mainstream press, report anything along these lines.   It isn’t selling newspapers (yet – watch this space).   Look at the BBC tech news (sorry BBC, I know I am down on you a lot) most days and its tech reporting is, at best, ridiculous (See this very morning’s:  Using AI for love letters seen as lazy, study finds.)

Now, I don’t want to be a doomsayer. I am fully invested in AI. I use it on a daily basis, and if I’m working eight hours a day, I’m using AI for seven and a half of them. It is a constant companion with me, working like a little Trojan, whether it’s ChatGPT on voice mode (in fact, this blog was dictated to CGPT, because I can talk quicker than I can type)), a code window coding for me, or the many other ways that I use AI, I am fully invested.

However, we are in a very interesting period where AI is coming at a speed that even I consider to be worrying. And what should make people pause is this: so many of the experts in AI, the head honchos in the companies that are running this, are sounding alarm bells. Only this week, Anthropic’s head of safety quite saying the world in peril, An article by Matt Shumar calles ‘something big is happening’ came out on LinkedIn and went viral and new LLM models were released – more on that later. The people building this technology are openly saying that it is accelerating fast, many agree too fast.

So what are the dangers of AI? They fall into two camps. Firstly, how it’s going to affect our business and working world. Secondly, whether or not it’s going to destroy us as a race. The second one isn’t necessarily science fiction, but it’s definitely out of the remit of this article. If you want to put the wind up yourself, just put in “dangers of AI” into YouTube and there will be an army of bits and bobs pointing out all the things that might go wrong.

I want to focus on business.Because for the business world, we are rapidly heading towards the time where AI can do more in most white-collar jobs than humans can. When I go to a networking meeting, I tend to look around the room and think, I wonder who in this networking meeting is in a job or running a company that will no longer be necessary in a very short space of time. AI can either do the job for them, or it has brought the price of purchase down so much and made it so easy for the general population to self-serve that trying to run a business in that space becomes very difficult.

And I hold myself as one of the first people up against that wall.

I sit in front of a computer and I code. In the last few months, I’ve been coding a little bit and also instructing AI to do my coding for me. At the moment, it probably still needs someone like me to think like a coder, think like a project manager, know exactly what to tell it to do so it provides the right results, and then check its work. But that is not going to be something that happens forever.

One of the new models that came out this week was ChatGPT’s Codex 5.3. I won’t get too technical, but effectively Codex 5.3 is a programming large language model that you can plug into your development environment, connect to your files, and get it to write code for you. Previously, with earlier versions, I would ask it to do something, it would do the thing, it would break my code, I would then ask it to debug, and we would go back and forth until between it and I things were solved.

Yesterday, I gave it a web application to build itself. Not massively complex (for the geeks here, it was an army builder for my wargaming hobby that lets me create an army list I can field on the table). I gave it almost no specifics. Within five minutes it had built it, working, without any further input from me. The prototype was up and running within a single prompt. It wrote all the files and did everything else.  That is a major leap forward.

However, what’s more important is the threshold it crossed.  In ai, we talk about thresholds points of no return that bring us closer to AGI – Artificial General Intelligence.  That’s when AI can do everything we can. Cheaper and faster, so that businesses use that instead of us. 

Codex 5.3 is one of the first large language models built partly by other large language models. I’ll say that again. The large language model just built a large language model. That puts us onto the next acceleration curve, because when it stops needing humans to build LLMs it will build them faster and better.

When that happens, so many jobs are going to be not worth doing anymore. We were naive, perhaps, when we thought it would be warehouse workers in Amazon being replaced by robots. We thought it would be blue-collar jobs. In practice, anyone who sits in front of a computer is likely to be in a situation where the things they do can be taken over by AI.

Is this the end of the world? Not necessarily. AI is very good at doing tasks, and just like I use it to do many tasks for me, I do not use it to do my job entirely. But, using it frees me to do more stuff that I could without it.

So what is the advice? Be an early adopter. The way I use AI is considerably different from most people who are probably asking ChatGPT a few things on their phone and using it as a search engine. As an early adopter, you can be head and shoulders above others in your field. Am I going to use the solicitor who does everything manually, or the solicitor who can do everything manually but  also uses AI research tools specifically built for solicitors? Adopt what comes along early and find the niches where you can use AI in your own business to put you ahead of your competitors.

That said, my own policy is this. I am running as fast towards AI as I possibly can, and I am also running as far away from it as I possibly can simultaneously. In addition to starting the Sea-Change AI brand this year and jumping in with both feet, I have taken a course to become a professional celebrant, looking after people’s weddings and funerals. One area I am convinced will not be taken over by AI in the immediate future is the spiritual, the emotional, the people first. I also did a plumbing course last year, which I hated. I am writing my own set of wargaming rules (sorry, geek again). I am convinced that as people have fewer work-based things to do, there will be growth in giving people meaningful and joyful activities to occupy them. The leisure industry will grow because people will simply have more time.

So if I can leave you with anything, it is this.

Run towards AI and learn it properly. But also build parts of your life and your skill set that do not rely entirely on sitting in front of a screen, because the screen is becoming far more capable than most of us realise.