Blog

I’m not a fan of the new image generation on ChatGPT. Not because it’s more detailed or more refined – the quality has gone up … but the fun images that I was able to generate for my posts quickly don’t seem to exist anymore. The new images take several times longer to generate, but I’ll admit they are much higher quality.

I also tried out the new video creation on GenSpark from a prompt from one of my previous posts. Whilst its a low quality video, you can clearly see a future where we can use AI to construct our environments visually in ways that have never been possible before. This is the video I generated below.

So fifty days have passed since I made a commitment to building ‘something’ with AI. I have roughly six months until a self-imposed deadline at which point I will properly reassess my situation. I am totally aware that the industry is incredibly competitive, and I’ve realised that more than ever, the real route to success is marketing.

But for marketing to be effective, and not to be run over by OpenAI … you need a specific niche, sector, problem or new value proposition because you can then use highly targeted marketing and lead generation for your product. If you have a general purpose product, you will get lost in the noise. So specific problems taken care of, mean you can target your marketing and sales. You can also tailor your product.

I didn’t initially have a specific idea … but having worked with SMEs for my entire career, I do know what business owners are looking for. There are already infinite tools out there but I wanted to build something in my own way and own style. It sounds cheesy, but web software has always been my ‘passion’ so this was more about one final attempt to take everything I’ve learnt so far and put it into one system.

I have begun a custom objects platform that I’ve wanted to build for a while – which I think can form the digital infrastructure for several niches, and ultimately the long term goal is to build my own flavour of AI assistant. Whilst this may seem wrapped up already, my experience tells me to build something that’s going to be useful for myself and people around me, and evolve it as I learn about the requirements that crop up as a result of pushing the AI agency. As the industry quickly evolves, it’s mostly going to be about persistence and spotting opportunities.

Aside from my own product, I’ve been exploring a partnership with an AI platform focused on the online gambling sector, starting the foundations of an AI agency to see what that entails and now more recently I’ve taken on board an online booking system transformation. This is all on top of jumping into the AI R&D scene, learning about the various techs.

So I’ve been quite busy, and it’s an enjoyable life really. I much prefer it to being a developer with no real impact or choice about what they get to work on. I’m aware I need to make money with my product at some point, but for the moment I’ve got a few steps left on the platform that I want to get done to get it to a point where I can apply it to different use cases.

The main thing I have learnt about in the first 50 days is how powerful LinkedIn can be if you were to work at it. I haven’t had too many likes on my posts, which I never really expected anyway since I started off with less than 100 contacts and the content was more for personal motivation than anyone else … but I have had some key things come up from it … people from my past have messaged me, and I’ve been able to reconnect to them and talk about what i’m doing. And when I want to a local conference, I was surprised by how many people were aware that I was at least doing ‘something AI’ related.

The point is, the last 50 days has made me realise that if you work LinkedIn properly you can always find work wherever you are in the world, regardless of industry.

Our brains are fundamentally pattern-recognition engines. Evolved for survival, they identify regularities in our environment to make better predictions and decisions. Learning is a process of repetition and refinement: every time we encounter a situation, we subtly recalibrate, inching toward optimal responses. It’s like an organic algorithm optimizing itself with each iteration.

So basically, part of our intelligence is how we recognise patterns as we go through life. It was a matter of survival many years back. Eat that red berry and you get sick, stamp on that snake and it’ll bite you… but more than that … what about sales psychology … use these evocative words, anchoring, mirroring, and positive affirmation statements … and you are more likely to get more sales. So patterns have already been assessed.

Science is literally built on patterns. The scientific method is about repeatable and measurable tests. You look for patterns in data to assess whether an hypothesis is correct or not.

So …

This idea leads naturally into how Large Language Models (LLMs) work. At their core, LLMs are also pattern recognizers. They don’t understand meaning the way we do — instead, they analyze massive amounts of text and learn to predict the next word based on patterns they’ve seen before. They don’t “know” facts; they generate statistically likely sequences. Think of them as highly advanced autocomplete systems trained on a trillion examples.

Whilst LLMs aren’t fully intelligent… if we were ever going to reach genuine AI … this pattern matching ability is very much part of that. Who knows what will happen with the quantum chips… since what does ‘meaning’ mean anyway? AI can ‘know’ everything about a dog, what it looks like, sounds like, how it behaves, etc… maybe it can’t ‘experience’ a dog … but it certainly conceive of one.

This gives rise to the thought that we have created something potentially very different to us, but that is still intelligent but in its own way. Our human ego don’t want to admit that this ‘thing’ may have more intelligence, so we have that internal roadblock… but why are we so fixated on it achieving ‘intelligence’ anyway – AI doesn’t need to be intelligent to be self-organising.

While LLMs are masters of mimicry, human intelligence is something deeper. Intelligence isn’t just about spotting patterns — it’s about applying them to adapt, solve problems, and pursue goals in unpredictable environments. It includes:

• Perception (pattern recognition)
• Learning (memory)
• Reasoning (applying knowledge flexibly)
• Agency (goal-directed behavior)
• Generalization (using insights across domains)

LLMs replicate the appearance of intelligence — eloquent, insightful, even persuasive — but they lack goals, memory of past interactions, and any sense of self or purpose. They are mirrors of the data they’ve consumed, not minds of their own.

It can easily be argued, and has been by philosophers over the years, that the human animal is nothing more than a biological robot… that many of us go through life without ever truly thinking; instead just riding the waves of the mind; which of course is just made up of what it has learnt over time.

True intelligence, especially human intelligence, combines cognition with emotion, instincts, and an internal drive. It adapts to change, grows from failure, and learns not just how to do something — but why it matters.

So what is intelligence? It’s not just knowing things. It’s the ability to use patterns in creative, adaptive, purposeful ways. LLMs are incredible tools. They’re not alive, and they’re not wise. But perhaps AI doesn’t need to be ‘alive’ and ‘intelligent’ to be an existential threat to us. I’m not saying LLMs are Terminators … but I am saying they are part of them. It just needs a few more technical breakthroughs combined with them, and then it’s time to reach for the EMP grenades.