Lately I’ve been dabbling and experimenting with an AI bot running on my laptop called Gemma. Mostly, I chat with it through a piece of software called Ollama, itself a sort of open source zoo for wild AI models. 

Gemma is one of a growing collection of multibillion parameter AI models that is available for free online to run on a home computer. These are part of the remnant detritus of big AI, bits of code have been set loose on the web, abandoned or “open sourced” by their creators. This is done under the guise of benevolence but  maybe it is more because they are not quite as clever and fancy as the latest and greatest flagship AI models running amok thru their data centres, and to which they can sell access.

Cory Doctorow noted near the end of his recent book The Reverse Centaur’s Guide to Life After AI^[1] that such freely available algorithms may turn out be one the only positive relics left behind when the (some suggest) inevitable collapse of this too-big-to-actually-survive industry eventually (maybe) happens. Reading the final chapters of his book may have led me to hunt down and install this software. The fall of big AI will leave a lot of mess and broken dreams but too it will too leave us software that will continue to compute on personal computers and do good work for small businesses and not-for-profits with shallow pockets—so long as folks like you and I choose to understand and run them.

Gemma is slow and a bit clunky and it is limited by the mediocre chip in my laptop computer, but it shows potential for being a fun little AI pet for simple chat and some lite automation, all with potentially, marginally, compartmentally less guilt of lapping up those steep data centre morality costs. Patience is a smaller and smarter price to pay than destroying ecosystems. I mean, it still uses electricity but… presumably much less.  And I’ve yet to divert a small waterway to cool my laptop.

I’ve been thinking about this moral problem as I write here and—to be honest—I remain conflicted. Being completely anti-AI isn’t the answer. Yeah, we should be wary. Yeah, understanding that the firms that create these things have gone astray and abused financial and political systems to enrich shitty folks with lacklustre ethics is definitely a take in the right direction. Yeah, there has been a jaw-dropping environmental cost paid to get to where we are. But being mad at a useful piece of software and learning how to leverage it for good things is a choice, too, and one we might all need to wrestle with in the near future.

Being purposeful and honest about responsible use of this new tech, well, that might be the most human part of this whole thing.