AI - The horror, the horror

Anyone who is taking a look at AI cannot help but be amazed by the speed of change just this year alone. For many of us, the first we became aware of a significant development in AI was with the release of chatGPT in 2022. I'm not sure where I first read about it, possibly on something like Twitter but I signed up in November 2022, had a quick play, and thought, over-hyped. Even then there were a lot of people saying how it was going to change the world, as is usually the case when they have a vested interest.

Had a second proper play in Jan 2023 and shortly afterwards a friend of mine told me about Synthesia and similar products and it dawned on me that there may be something to this, that this thing might be a game changer. I also realised almost immediately that there were issues on the way.

I started web development back in 1996, yes, that is almost 30 years ago, and despite the rewrite of history, one of the real driving forces behind the web uptake was porn. The porn sites introduced the world to something that was known at the time as the JavaScript Jerk Around, at least this was what it was known as in London where I worked. This was the use of multiple pop-ups, pop-unders, repositioning, resizing, and closing prevention functionality delivered using JavaScript. This type of aggressive technique was one of the reasons that the more regulated web came into being, placing limits on what websites can and cannot do without user consent.

No one is completely sure what the AI equivalent of the JavaScript Jerk Around is at the moment, but you can be sure it is coming. OpenAI has held off on the general release of Sora and audio functionality that makes it possible to replicate someone's voice from as little as 15 seconds of test data. Why? Lots of reasons, but one of the more obvious is to stop people from doing something like this...

https://twitter.com/nomadjourney33/status/1773116360547041727

This isn't Donald Trump talking and might be made from "traditional" methods of editing, but with AI something like this is going to be super easy for anyone to produce. However, this is just the tip of the iceberg. Imagine a video supposedly from the sage of Omaha, Warren Buffet, giving something that seems like genuine financial advice. Imagine the video format used is similar to how Warren Buffet does it now. People may take it to be sound financial advice and only later discover that it is a deep fake and that they have been the victim of the upgraded Nigerian 419 fraud. Think this is a little too sci-fi, have a read of this, it happened for real a couple of months back

https://edition.cnn.com/2024/02/04/asia/deepfake-cfo-scam-hong-kong-intl-hnk/index.html

So where does this leave us today? The genie is not going back into the bottle that is for sure. Huggingface has almost 600,000 models available for download at the moment. I have about 50 of them on my own PC. Even if Huggingface disappeared today and all the big companies decided to pull AI for the time being, me and millions of others could still use the models we have. If you are reading this a year from now, click on the link here to see how many model they currently have,

https://huggingface.co/models

I expect it to have topped a million. Many of these models can be used as starting points for AI development. These models cover such a wide range it is incredible.

At the moment there are clouds on the horizon and the AI storm is heading our way. It isn't going to be good that is for sure. For me, it is not all doom and gloom and if anything, it has helped solidify my own thinking. AI is a truly game-changing technology, I have been looking at education for a couple of years now, and with the advent of AI we have a real opportunity to deliver a significant groundbreaking revolution in education and I will be going into this in greater detail another day.

Immediately, right now today, we can use AI within corporations, businesses, and any organisation for significant change. Yes, the same worries about AI abuse exist within a corporation et al, but the significant difference is that if you catch an employee using AI to create deep fakes, or anything else untoward, you have the option to fire their arses and prosecute them to the full extend of the law should you decide to do so. This threat already exists and it stops people from crossing the line every day in many many jobs. Getting fired for doing something illegal is often deterrent enough for most people.

What this means for us is that we can roll out AI within corporations and businesses today. We can start taking advantage of AI within any organisation right now. We can start training employees in the use of AI today. It is more than this though, it should not be "we can", it should be "we are".

Firms should be training their staff in the use of AI right now. They should have their IT people preparing the way for AI deployment across the enterprise today. In a post a while back I mentioned that every job I have worked on over the last 25 years would have benefitted from the use of AI. Some of this would have been helping the developers to write better code and to produce better testing scripts. Some of it would have been the use of AI within the company to improve efficiency and to open doors that were simply closed due to cost and time.

I'll give you an example. I worked for a telco a while back and there were a series of reports that would have been "nice to have". They were reports that would probably have been helpful to the firm, but it was hard if not impossible to determine the ROI for these reports. Having never had the reports it was difficult to justify spending many tens of thousands and spending many months developing the software to deliver the reports because no one was sure if they would prove any good. They were a "nice to have", rather than a "must have". In many jobs, "nice to have" never gets delivered. What if you could now get those reports for a fraction of the cost in a fraction of the time?

What if you could use AI to summarize a 400-page document in a few minutes rather than having someone have to read the report and make that summary? A job that might take them days or even weeks. These and many other things are available to businesses right now with existing AI available today.

In the coming posts, I am going to focus on how AI can be used by businesses today. Over the years I have worked for telcos, the NHS, government, financial institutions, multinational corporations, law firms, digital agencies, and multimedia news organizations. All of these can and will benefit from AI.

I thought I would start posting on how AI could have been used within the development process and also what AI can deliver for these firms. Much of what I did over the years is still relevant today. Corporate intranets and the functionality they deliver still serve the same core purpose.

I'll try and do two posts a week and we shall see how we go. That is enough for today, I'm off to get the AI to write some more code for me to review.

Bryan
ps. This continues to be written by a human, me, and summarized by an AI of my choice, see the summary for details.