How do I keep up with everything AI?

This short article is written for both newbies and those already familiar with AI models who are struggling to keep up with the rapid pace of new AI models and tool releases.

It’s been almost a year since I last posted and for those that know me it was an awful year with some major personal losses, resulting in a change in my perspective on life. I needed time to step back and focus on the things that mattered. Nonetheless, for the past year, I have still been keeping myself in the loop (pun-intended) by loosely following some of the major developments in the AI/ML and tech space. To be honest, it’s getting increasingly difficult to do nowadays, unless you kind of know what you’re looking for. It was already hard about 7/8 years ago with the surge of deep learning research papers; however, we’re now seeing papers come out at an unprecedented pace. The democratisation of the AI/ML space has led all sorts of players (both good and bad) to get involved and while we are seeing some incredible research and AI products come to the market, we still need to sift through a lot of noise.

So how do I stay updated? Well, it’s simple – I tend to not look at every release right away. Not only would this lead to burn out, but it’s impossible to keep up with everything. I have a gentle rule of thumb on how to navigate through the hype, and it mainly relies on one thing: waiting. I will typically wait at least a month before I read a newly released publication or download and try the latest app – especially if the premium version is where the real utility lies. I may stumble across a post on social media or read a review of an AI model or new tool, but I still won’t read the newly released 55-page publication. Often, these papers contain no real insights and can be (IMO) marketing material guised as serious scientific research. By simply waiting a while I can ignore hype nonsense like “10 WAYS THIS CHANGES EVERYTHING!!!” or “Why blah blah blah release means you will lose your job”. If a new model or tool is still trending or is considered worthy of discussion after this time, only then it will be worth the time and money to explore further.

It’s ok not to be fully on top of the latest in this space as most of the new releases are minor incremental gains or worse, just hype. And for those who are new to the space, these days most of the AI discourse is on generative AI and Large Language Models (LLMs) as seen with ChatGPT, Gemini, LlaMA models and Claude. In my opinion it would be better to gain an understanding on how these models work and what it takes to train them. This doesn’t need to be extremely detailed, and I would even encourage just attaining a high-level understanding. It is only then you can appreciate how much effort it takes to create an AI/ML model and why it is really difficult to get right.

If you have an hour to spare, I recommend watching one of the best introduction videos on LLMs by Andrej Kaparthy:

And this excellent article by Wolfram: https://writings.stephenwolfram.com/2023/02/what-is-chatgpt-doing-and-why-does-it-work/

Once you have developed a better understanding of AI models, you are better prepared to cut through the grift and hype as your healthy scepticism becomes your first line of defence.

So, How do I keep up with everything AI? Well… I don’t. I wait a month (or so) and avoid some unnecessary painful reading and/or wasted money on apps and tools that might be considered useless later.