The claws are open, until they close around you...

Published on: 2026-02-18

...and they may well not be under your control

Pretext

I recently changed my job. Why? It made sense with a young family to try and cut the amount of travelling I was doing. I don't fly, but being away from them and burning a lot of diesel in the process made it a bit silly. Especially when you consider that I'm predominantly an experienced software engineer with a penchant for enabling others to compute, and compute hard.

Openclaw is the flavour of the moment and I first came across it through Moltbook. The number of name changes to OpenClaw is indicative of it's ridiculous development velocity (I know of three, yes?) Given that it can't be more than a year old, the impact of this project is astounding.

Let's be clear, with what I now say, the ingenuity of the invention is in no doubt. This is machine learning (don't get me started on how it doesn't represent all AI) finally working hard.

Agentic AI unleashed in this way is a terrible idea.

Why bring this up?

With the sheer level of integration possibilities into end user systems, we're allowing unproven and poorly engineered bullshit generators to integrate with our lives. Given the spread of misinformation globally, this poses an information nightmare for the future generations, whom we owe a responsibility to. If we're employing technology, there needs to be an end goal in sight that discernibly makes things better than we found them.

Environmental concerns

Everything takes power to run and these agents are no less power hungry than, well, something that could be far more useful. Rather than letting them loose as an exercise in seeing what they can do, perhaps isolate and experiment with them responsibly. I'm a hypocrite at this point, because I've acknowledged the scale of the problem with others. However, we stated that the first thing is to be aware of the cost of what you're doing. That's why I'm working with what I do today.

Ethical concerns

Therein lies the ethical rub. Why are we facilitating integration without considering impacts such as time-wasting for individuals in the real world? We are wasting the worlds resources with these highly experimental bullshit generators before extracting the usefulness from the technology we already have to hand.

And there are great uses for deep learning based technologies, among others. I've been very privileged to be involved with many projects that exemplify the potential.

But it's still only potential, and we are only just finding our feet. We must use technology responsibly and cannot absolve ourselves of the responsibility of judging each and every action we take as technologists, because each joule of energy and millilitre of water taken up by these technologies is taken away from someone else.

I have intentionally tried to tone down the criticism of the AI hype bubble which quite frankly is a load of 40-somethings deluded in their faith in themselves. This constant pursuit of agentic AI being a gamechanger without goals was really nicely highlighted as a concern in a recent interview with Meredith Whittaker of Signal. Kudos to her.