Another month, another deepish dive by the three podnosticators of the SmallDataForum – who Sam describes as “Thomas = the philosopher-academic and historical context-setter; Neville = the champion experimenter and enthusiastic evangelist; and Sam = the dabbler, observer, and sceptic.”
This time, we dive into generative (as well as degenerative) artificial intelligence, large language models (LLMs) and various chat-botty applications, including Neville’s new favourite, Claude, “the most human-like experience”. Turing Test, anyone?
Perfectly timed with our latest podcast release, Quadriga University Berlin launched an e-book on AI and PR, edited by friend of the show, Professor Ana Adi.
Artificial Intelligence in Public Relations and Communications: Cases, Reflections, and Predictions contains timely, critical, insightful essays from practitioners and academics. This includes a piece by yours truly, informed by decades of stochastic (a posher word for ‘random’) knowledge acquisition.
In keeping with the role of historical context-setter, I link the modernist engineering mindset of AI to the beginnings of modern capitalism and the public relations industry (see Forbes article on AI and social engineering). Goldsmiths communication scholar Clea Bourne makes a similar point – though less focused on history, and more on presence and future – in a 2019 paper, as well as in her contribution to the AI in PR e-book.
The PR industry, she argues, acts as AI cheerleaders where critical reflection and questioning is required.
Sam agrees and adds his observation regarding the lack of intellectual heft in the PR industry, which after 15 years in the trade made him look for smarter questions and smarter answers elsewhere – in this case, a PhD in experimental-behavioural psychology.
Skim-reading the Quadriga piece confirms his view that there are degrees of sophistication between the German PR industry, and what counts as high level discourse in the UK and the US on the difference between psychology and behaviour change.
Neville reminds us of his reflections on PR’s having its head in the sand on AI, ostrich-style, in a March blog post. He muses that this is not the first time that PR is caught out by technological progress: he recalls the halcyon days of desktop publishing (remember Quark Xpress?), when the presence of an Apple computer sent whole account teams into a panicky flap. It’s all about control, see – of message and of (production and distribution) medium.
While Neville generously compliments my academic angles, insights and sources, I wonder about my ‘bubble opinions’: it’s all very well to talk of markets as discourses and social constructions, but when you’re being hit but a cost of living crisis, I suppose you care more about practicalities, than social science lingo…
Sam points our attention to the latest issue of the Harvard Business Review: Reskilling in the Age of AI (hot on the heels of the previous issue, Gen AI and the New Age of Human Creativity). Has it really already been eleven years since the HBR declared data scientist to be The Sexiest Job of the 21st Century?
He questions whether AI really has ‘hit the mainstream in record time’ and cautions that we need to pick up signals beyond the likes of Gartner, BCG, Deloitte and McKinsey – who are mainly in this for themselves, as Mariana Mazzucato and Rosie Collington demonstrate masterfully in The Big Con. It is hard to escape the hype, Sam says, with LLMs as AI’s poster boy, and ChatGPT as the poster boy’s poster boy.
Neville agrees but also points to AI applications beyond the current hype. When AI chatbots can operate software businesses with minimal costs and minimal intervention (as in this Business Insider example), there are plenty of big and small business applications. And as Sam rightly says, with the capabilities of ChatGPT and similar, we’re all coders now.
There is breakout technology, Neville states, and there is fear-mongering and hype-mongering clickbait. The problem is our skewed awareness landscape. We might add skewed expectation landscape to that…
While Neville takes us through some of the highlights of McKinsey’s The State of AI in 2023 report (executive summary: you REALLY wouldn’t want to miss out…), I’m referring to some numbers I’ve come across in Clea Bourne’s chapter on AI Hype and Public Relations (in the aforementioned ebook). So we are talking about what really makes the world of AI go round – and it’s not data…
Following some doomsday comments in tech investor newsletters, “the US stock market had a resounding rally of tech stocks, with more than US$4 trillion added to the value of the NASDAQ 100, while the S&P 500 surged 159%”.
Together with regular headlines about ChatGPT’s record-breaking adoption, and breathless predictions such as PwC’s forecast of a potential US$15.7 trillion AI contribution to the global economy, and up to 26% boost in GDP for local economies from AI by 2030, what we are experiencing is numberwanging with real consequences. Or as previously mentioned: markets really are just discourses.
We move on to an Ethan Zuckerman piece about the problem with AI training itself, and the fact that sooner or later, we will run out of distinct human-generated output. Neville briefly discusses an interesting scientific paper mentioned in the article: The Curse of Recursion argues that training AIs on their own outputs would be like “engines choking on their own exhaust” – the end result might be model collapse.
As with pretty much everything in AI, it is probably ‘too early to say’.
Which is just more reason to approach with caution and critical thinking. The ChatGPT interface has a disclaimer that states “ChatGPT may produce inaccurate information about people, places, or facts”. So, about everything, then.
Sam is reminded of the Hollywood writers’ strike and the role of AI. A recent FT article discusses the phenomenon of the ‘uncanny valley’, when AI outputs and robots humans will really become indistinguishable from humans. Earlier this year, The Guardian asked “It’s 2023, where are the sex robots?” Apparently, we are a long way from realistic, real-life experiences.
But even if McKinsey’s State of AI study doesn’t touch on this significant growth opportunity, the podnosticators will keep an eye on developments…
Listen to Episode 77: