Always with fingers on the pulses of the most relevant breaking news stories, the not yet scurvy-plagued triumvirate of the SmallDataForum briefly contemplates the shortages of fruit and veg on Great British supermarket shelves.
And we decide that neither the Marie Antoinette-esque “let them eat turnips” intervention of political-turnip-made Secretary of State for Environment, Food and Rural Affairs, Therese-with-accents-aigu-et-grave Coffey – nor the seemingly permanently unflushable turds, former-now-shadow Prime Ministers Johnson and Truss, are topics worth any of our (or our listeners) attention.
Sam, of course, wouldn’t know much about those domestic five-a-day-struggles, given his jetting all over EUlandia (Catalonia, Amsterdam etc.), promoting his excellent, not-to-be-missed Using Data Smarter online course, building a “digital ecosystem” – and ZING, just like that, Sam won this episode’s jargon bingo.
How he finds the time to read Times columns is beyond me. But read he does, and so we find ourselves discussing David Aaronovitch’s piece Nobody wants what the Tories are selling (if only they were selling fruit & veg).
From the Ditchley Park secret Brexit management summit, to zombie Prime Ministers, to remarkable polling figures (“In a fairly typical YouGov poll from the turn of the year, 2 per cent of those aged 18 to 24 supported the Conservatives”), Aaronovitch’s assessment of this “balloon of mediocrity kept aloft by hedge fund billionaires” oozes the grumpy sophisticated ennui of the lefty middle-aged male British journalist-commentator – echoed masterfully by Jay Rayner during his very recent appearance on the News Agents podcast.
We muse over future election results and whether or not a Labour landslide – a hefty 27 years on from the last one – is a certainty. Sam thinks so, Neville isn’t quite so sure. But perhaps the real question is – would much change under Keir “five missions” Starmer? It’s fair to say we don’t muster a great deal of optimism between us.
My concern is that, since there isn’t much else left for the Great Con Party, their strategic focus – directed by ‘mastermind’ Isaac Levido – will be on waging an ever more frantic (desperate?) culture war, with such able battle field commanders as Suella “stop the invasion” Braverman, Kemi “the smart face of culture-warring bigotry” Badenoch, or Lee “off with their heads” Anderson. What could possibly go right?
From culture war bots, we turn our attention to real chatbots and other generative AI-powered wonders. As the prime tech enthusiast among the three of us, Neville has done the most exploration of new AI tools over the last year or so. ChatGPT in particular continues to impress.
With all due caveats – its early days, it needs a lot double-checking, hilarious stories of weird results abound – Neville finds it “outstandingly good”. Despite some remarkable misfiring (see example below), by and large I agree.
No surprise, then, that both Neville and I are early ChatGPT PLUS adopters, at a tolerable $20/month. Paying for an experimental online application – O brave new world. Even Sam hasn’t been as excited about the Internet “since about 1995”, even though his focus is currently more on digital management and marketing tools to build his aforementioned digital ecosphere.
The question of whether, how and when we’re witnessing a major disruption of the Interweb’s cherished ad-funded business model of “if you don’t pay the product, you ARE the product”, toward a more traditional “paying customer is king” approach, needs further and deeper exploration (Andalusia trip and Rioja-fuelled deep-dive podcast recording, here we come).
At the same time, Microsoft is planning to serve ads in ChatGPT powered Bing chatbot results. However this all is going to play out – will Bing beat the Bard? Does the emperor of the metaverse wear any clothes? – there will be a lot for us to dissect, analyse and evaluate over the coming episodes.
As for my own practical experience with ChatGPT’s wrinkles, I recently tested its academic writing mettle with this query: “Write 200 words on the risk society with academic sources, verbatim quotes and bibliographic references.” It came back with multiple Ulrich Beck references (so far so good – he wrote THE Book and several others on the risk society), plus this: “According to Hulme (2019), “The most disadvantaged populations are most vulnerable to climate change and are least able to cope with the impacts” (p. 7). It also provided the proper bibliographic reference:
Hulme, M. (2019). Is the concept of a ‘risk society’ useful for understanding climate change? Social Epistemology, 33(1), 1-10.
The problem? Mike Hulme is a real Cambridge don who wrote a highly insightful book about climate change discourse. Social Epistemology is a real peer-reviewed journal. The referenced article, however, is not real. Doesn’t exist.
ChatGPT invented it.
It has all the hallmarks of an academic reference, only that it isn’t. It’s merely a ‘plausible hallucination’. And the more insidious for its plausibility. Perhaps that’s also a good thing as it shows that “ChatGPT PhDs” are really a long way off.
On the other hand, when handled with critical care, it can be a powerful and useful academic assistant, as Sam’s son’s experience with the chatbot’s summarizing capabilities shows. As pretty much everything in the long history of human invention, this may be used for good, or for not-so-good. The devil is in the moral compass of the user.
As briefly alluded to – the three perhaps wise and certainly wizened men of the SDF are finally off on their repeatedly COVID-delayed sojourn to the lands of Al-Andalus, for a long weekend in March of high-intensity podding and plodding.
The fruits of our Spanish inquisitions will be available here in due time.
Listen to episode 66: