Senko Rašić

Increasingly I notice when I talk about AI with others, we often mean subtly different things.

When I'm talking about AI, I'm talking about the technology (LLMs, agentic systems, etc.) and the type of product one can build using that technology (chatbots, assistants, classifiers, process automation, etc.)

I now see that for many others, that's not the primay concern. Instead they think about a class of (consumer) products (ChatGPT, WhatsApp AI chatbot, MS Copilot) and/or the effects of people (mis?)using the technology (AI slop).

To illustrate: In a recent conversation I stated I believed the AI models will improve in the future. In my mind, an LLM improving is an objective fact: I can task it with something and get better result.

The reply was “in which direction, and for whom?” which is a (totally fair) product and business strategy issue – not a technical one. We indeed might have much better LLMs within much shittier chatbots!

Another conversation from a few weeks ago was about AI serving porn. LLMs are gigantic autocomplete machines, they'll serve whatever you want them to serve (to a first approximation). From a technical standpoint, that's a complete non-issue.

Looking at AI as a class of products increasingly used by everyone, including our children, and including people who are not tech (or AI) savvy – what, how, and why the product makers choose to serve IS very important!

This is similar to “social networks”. What we today call “social networks” are everyhing but – a far cry from the “web2.0” social networks era, when the point was to interact with your (real) social circle. The name stuck, the damage is still being assessed (witness recent Australia ban for social networks for under 16s), but the “social network” aspect is not the problem: the hyper-optimized engagement machine peddling all sorts of questionable stuff is.

Sadly, I think the meme battle is already lost: AI is increasingly being understood as a class of products. The problem is when this mixup causes people to blame the technology for the questionable business and product strategies that big tech companies use to maximize shareholder value.

AI, the technology, is now blamed for layoffs, struggling artists, and slop, to name a few. In fact, companies have a great scapegoat for layoffs, Disney just made $1B from AI (none of which will go to struggling artists), and people have been hand-crafting slop in the name of SEO for years.

I don't have an answer. This post is mainly for my friends, to explain that when I'm bullish about AI, I'm bullish about the underlying technology. I'm not bullish about the way big tech is going to (ab)use it to maximize profit.

Hacker ethic is about using tech in inventive ways to improve people's lives. I believe AI has the potential to do so. Sadly, I know it will also be used by those of “Greed is good” ethic.

I hope when we direct our critique, we aim at the right target.