The last few months have seen an avalanche of AI-related activity in academic publishing. AI themed sessions filled the stages at the Frankfurt Book Fair. eLife hosted an open-source community call on the latest developments in AI and machine learning tools for open science. Atypon launched a new AI-driven app, Scitrus, which gives researchers a personalized, social media-style feed of only the latest and most relevant journal articles, news, and citation alerts. Unsilo published the results of their survey of academic publishers which included questions about their use and attitudes towards AI technologies. Frankfurter Buchmesse and Gould Finch published White Paper looking at the impact AI will have on the publishing industry as a whole. More cautious voices have started to express concern, for example, David Beer ponders what will be lost if researchers use AI to make themselves quicker and more efficient.
So much of the AI in publishing discussion focuses on worthy (but somewhat dull) process efficiencies that I was really excited to attend Byte The Book’s event about AI in the creative industries which promised something a little different. I wasn’t disappointed! Justine Solomons brought together an eclectic panel which included fashion director Candice Fragis, author Robert Elliott Smith, lawyer Alex Hardy and tech entrepreneur Taylan Kamis. Journalist Mark Piesing chaired the session and facilitated a wide-ranging and fascinating discussion.
Mark began by asking the panel to define ‘artificial intelligence’. Robert Elliott Smith, whose new novel Rage Inside The Machine challenges the assumption that technology is an apolitical and amoral force, noted that what we mean by ‘artificial intelligence’ is a moving target. What we call artificial intelligence today is very different from what it meant in the Seventies. Robert, borrowing from Neal Stephenson’s The Diamond Age, prefers the term ‘pseudo-intelligence’ because it differentiates AI from human intelligence.
On the topic of AI replacing people, the panel seemed more upbeat than many commentators. Taylan Kamis from DeepZen, a producer and co-publisher of machine-read audiobooks, has chosen to base his service on the voices of real actors rather than computer-generated voices with the actors receive ongoing royalties for their cloned voices. Candice Fragis, whose fashion app DREST claims to be the world’s first interactive luxury styling game, commented that machines don’t have a gut instinct, you need people to curate and inspire, to show the consumers the products they didn’t know they wanted. Didn’t know they wanted was the crucial bit here, algorithmic recommendation systems are getting very good at learning what we like and giving us more of the same which could be leading to feedback loops and filter bubbles.
Inevitably concerns about bias were raised “all avatars carry cultural bias of many 100s of years of human behaviour”. It’s not clear if this is a problem that can be solved. Will your avatar succeed if it doesn’t behave as society expects? There are reasons why most voice assistants default to female voices.
Alex Hardy, from the entertainment law firm Harbottle & Lewis, made a very interesting point about allowing AI to crawl copyrighted or subscription content. If you don’t allow your work to be indexed by an algorithm are you effectively taking your work out of the cultural conversation? It’s an interesting question for publishers, would we prefer the algorithms to learn from high-quality content that is often closed behind a paywall or copyright restrictions or would we prefer the machine to learn from content freely available on the internet that might be of questionable quality?
On the limits of AI in the creative industries, I think we’ll see many more applications and wider adoption than the panel. Although it’s still early days Healx’s scientific abstract generator hints at a future in which much scientific writing (which is often quite formulaic) is generated by a machine and polished by humans. Holographic concerts have been around for a while suggesting that a concert experience doesn’t always need human musicians. Candice Fragis commented that in Asia there are avatar influencers who have more followers than their human counterparts. Kids in Korea don’t discern between humans and avatars, they just see a style that appeals to them, and think “I want to look like that”.
One of the more thought-provoking closing points was made by Robert about AI technologies leading a ‘two-tier” system in the future. Borrowing again from Neal Stephenson’s The Diamond Age, the idea is that the rich will get expensive hand-built quality products whilst the poor will get cheap synthetic products. Sadly, it’s not too hard to imagine this happening. For example, if you can afford to pay you will be able to see the human doctor with the personal touch, if you can’t afford to pay you will get a basic service with the ‘robot’ doctor.
AI in the creative industries was a fantastic event which demonstrated the power of bringing together diverse thinkers and a diverse audience. You can join the Byte The Book network here.