Everywhere

In theory, and at least partly in practice, the fact that I’m on a two-year research fellowship means that I can adopt a somewhat laisser-faire, sufficient unto the day, mañana mañana attitude to aspects of the day job about which I’d normally be getting worked up. Curriculum review to transform all programmes into dynamic impactful future-skills-orientedness? By the time I have to teach again, the university might have worked out what, if anything, its buzzwords mean in practice. Use of GenAI in assessment now ubiquitous? Again, in a year’s time the university guidance might have settled down (and the one thing to be said about forward-looking curriculum review is that it surely makes a wholesale reversion to unseen exams a lot less likely), and I can work my way round that; hey, it’s even possible that the GenAI bubble might have burst by then, either because the financial model has collapsed or its limitations have become impossible to ignore. For the moment, I can blithely postpone worrying about it – which of course is one reason why this blog has been quiet recently.

But reality does keep intruding. Announcements of devastating job cuts in the humanities in general and classical studies and ancient history in particular – Cardiff, Macquarie – are horrendous in themselves, for the students and colleagues affected, but it’s also hard not to feel a chill global wind, a spreading international suspicion of forms of knowledge that aren’t immediately practical and vocational (leaving aside the fact that all the young people who were pushed towards Very Future-Orientated computer science suddenly face a much bleaker job market than your typical humanities graduate), that translates into the same consultant-driven senior management groupthink death spiral. It’s difficult not to listen to politicians and media types sorrowfully or gleefully concluding that the majority of young people need to be denied the opportunities they enjoyed – that the value of education needs to be maintained by making it a scarce commodity – and lurch between fury and despair.

And then there’s the GenAI stuff, simply because it is EVERYWHERE. Media stories about widespread ‘cheating’ in universities sit alongside stories about AI being the government’s big idea for economic development – suggesting that it’s a toss-up whether they are more ignorant about universities, students, the economy or the capabilities of LLMs. The injury of how useless and time-consuming Google web searches now are is compounded by the insult of the GenAI summary at the top, which I have to try to ignore because otherwise I get caught up in analysing exactly how and why it’s misleading or unhelpful, rather than spending the time trying to find search results that are less useless. “AI responses may contain mistakes”: indeed. None of this is good for the soul; I was already prone to assume that I know more than most people, and this arrogance is being compounded every time I try to look something up.

Whenever Google is unspeakably crap, Meta can always strive to be worse. Thankfully it seems to have disappeared again this morning, either because it was another of their testing-the-water things (like artificial users) or because I’ve ignored it sufficiently for it to go away, but last week, on the limited occasions I opened the Facebook app, every single post had a GenAI summary or a button underneath encouraging me to ask questions about related topics. This is simply infuriating. Increasingly I stick to my aged iPad, which hasn’t been capable of accepting updates for several years and so remains free of most of this rubbish, plus being by far the best platform for my bat detection gizmo (newer models, like my wife’s, seem to over-render the data in the visual display and then keep freezing) – and can’t manage Facebook at all. But I know that at some point the machine is going to die…

I don’t want to use GenAI, but it’s so pervasive that there’s a constant risk of clicking on the wrong button. I want to ignore it, but it’s everywhere. The corruption must be pervasive, the thing must be ubiquitous, so that people come to take it for granted and start using it enough to make the business model work – and if they don’t want it, it must be shoehorned in regardless (not least by preying on the gullibility of desperate-to-be-relevant politicians and management, and by spending a lot of money).

The big example last week of ‘We can’t have nice things, they must be enshittified’ was the iNaturalist app. This is a wonderful worldwide citizen science project, that allows you to upload geolocated observations of creatures and plants (with photos, or in the case of things like bats recordings of their ultrasound signals), either with your own suggested identification or using tools on the site to find suggestions, and other users then confirm the identification or offer alternatives. I’ve been uploading my bat observations for a few years, so it serves as an ongoing public record of bat life in the neighbourhood (and is the platform for the annual communal Big Bat Count that I organise), and there is a thrill in having an observation upgraded to ‘research quality’ because someone else has confirmed it. This is used as a serious research tool across the globe, as well as by lots of private individuals (including, allegedly, by the Australian mushroom poisoning woman as a means of finding localities with death cap mushrooms, but I don’t think that outweighs the positives).

And then last Monday they uploaded a blog post (https://www.inaturalist.org/blog/113184-inaturalist-receives-grant-to-improve-species-suggestions):

iNaturalist is excited to announce an award from Google.org Accelerator: Generative AI to help build tools to improve the identification experience for the iNaturalist community… Our nonprofit mission is to connect people to nature through technology and advance science and conservation. We see this new opportunity with Google.org as a clear extension of the work that we’ve been doing for years to build better tools to connect people to nature. By using generative AI (GenAI), we hope to synthesize information about how to distinguish different species and accurately convey that to iNaturalist users.

This was, to say the least, not popular among the sort of serious users who engage with their forum rather than just upload stuff, and the people who run the app have been hastily issuing caveats, insisting that user data is protected and that users will be properly involved in evaluating test projects kept completely separate from the main system and so forth – they’re not handing anything over to Google, just drawing on their expertise, honest.

Even if that’s entirely accurate, it still leaves a nasty taste. Why did anyone who’s an expert in this stuff consider it a good idea to get GenAI involved in discussions that require expertise and judgement – and an imperative, sometimes, to say that “we don’t know” is the most reasonable answer (distinguishing species of Myotis, for example), rather than generating nonsense in order to deliver an answer? Yes, they’ve used machine learning for years to help with identification – so do they not realise that GenAI isn’t the same thing? Is this another case like academia.edu, where something that looks like a public service is actually just collecting users and data until they reach the point where they can start to monetise them? In this case, apparently not, as iNaturalist is a non-profit organisation, but it still creates suspicion, and a feeling of having been much too trusting in imagining that a useful online platform and community might indeed be just what they seemed. How naive.

At the least, it offers a sense of how companies like Google are investing money in grants for organisations to experiment with GenAI, with the aim of identifying uses that are good and positive and not at all undermined by the tool’s innate flaws that can then be deployed to cloak the whole enterprise in virtue (cf. all the ‘AI improves cancer detection rates’ stories as if this has something to do with ChatGPT or Gemini), and to make it as ubiquitous as possible. And perhaps because they can’t bear to have any part of the Internet that hasn’t been touched and corrupted by their values and worldview.

I’m still going to use iNaturalist, at least until things go markedly downhill, because it offers something that isn’t available elsewhere; just as I continue to use Facebook, at least occasionally (regular Sunday chat with a group of friends and friends of friends). But it will now be with a definite edge of suspicion, wondering what the platform may be trying to get out of me or whether the data I’m giving it may be misused, and a sense that, if you go online at all, it’s pretty well impossible to avoid contamination.