There’s a lot of people who think AI is overhyped…
I was thinking about this today and realised that it’s complete rubbish.
Language Models completely skipped the ‘hype phase’ and went straight into skyrocketing user adoption phase.
To me, hype is what happens before something is ‘launched’ … like the hype of a movie or video game … big AAA titles get hyped up and then invariably fail spectacularly in the game industry. Movies that are just rewrites of old stories are also very culpable.
Crypto was hyped up for sure – albeit the main ability to have a citizen currency out of the control of third parties seems to have been missed in the minds of many in terms of importance.
Don’t people know it’s a good thing to have competition in currencies? Or that giving the ability for a government to freeze your bank account based on social meta data, is a frightening prospect happening very soon?
But with crypto there wasn’t a stable enough coin for them to be worthwhile as value exchange (yet). It became about lambo mooning rather than escaping from central banking tyranny. Beyond that, the use case of crypto for the average person isn’t there.
Language Models are being adopted by the ‘average person’ super quickly. Even if it’s just to have sounding wall for things that are on their mind.
So, with AI … there wasn’t really any hype as such – the user adoption proves that. Talking specifically about LLMs, ‘AI’ is so flexible that hundreds of millions (billions?) of people are using it in their own unique way.
Is it creating actual tangible business value?
Yes. If the businesses are applying it correctly.
Will it disrupt industries?
Yes it already is, and once people figure out the best way to leverage it, it will automate, streamline and enhance many businesses. They will do far more with less.
I think at the moment tech people and entrepreneurs are happy to try out all sorts of different tools- and willing to pay for them for a few months to see how they go.
I think LLMs are just the next tool in the overall arsenal of AI – computer vision, machine learning, deep learning,
The way forward.
Got some interesting things going on right now, but will write about them if they work out.
At the moment my biggest challenge is finding sufficient time everyday to R&D in the AI. I think the way forward is:
- I want to rent some dedicated server options where I can use the largest deepseek models … this is mostly just out of interest, but also it’s the first step toward proper privacy of data – I honestly think it’s insane that everyone is sharing their innermost thoughts with these AI companies, Silicon Valley doesn’t have the greatest of reps when it comes to privacy. But the point is language models can be private to you. So I want to investigate that, and work out pricing and performance; and also learn a bit about tweaking weighting of models and see how that looks.
- I want to experiment with much larger and detailed prompts – coding and normal – to see how far I can push them. For instance with Claude I have built mini Vue applications which were really good quality but they had to be properly steered by me. So a lot more experimentation on how good I can push these things.
- I want to learn and experiment more about the image, music and video generation. GPU servers are slightly more expensive for me at the moment, but I’m hoping a friend is going to help me out on that. And I want to make more fun cat videos.
- I’d like to see how much it would take to build some sort of mini-ITX form of computer with sufficient enough RAM to run language models locally.
- I’d like to experiment with micro models to see how they perform on mobiles.
- I want to experiment with scaled sentiment analysis
- I have made a custom object framework infrastructure and I want to link that in with web crawling and language models
- I want to experiment with the ‘talk-to-your-data’ tools
- I want to experiment with the automation tools (make, n8n)
- Lots more web crawling transforming into structured data sets of my own i.e. knowledge cubes
- Pushing the limits on frontend webdesign to see how creative I can be now
- Look into natural speech tech
- Bring all of these things together in one product
- Work out what stocks to invest in
Anyway more again soon
Leave a Reply