The 150th Block: Science adapts to generative AI, legacy media adapts to social media
Read to the end for a piece on nudge theory (TV series cancellation trend)
This week…
I was travelling for most of the week so I wasn’t too connected to what’s going on in the world.
Anyway, here’s a selection of top stories on my radar, a few personal recommendations, and the chart of the week.
Thanks to generative AI, catching fraud science is going to be this much harder
Katyanna Quach for The Register:
Image manipulation is already a top concern for academic publishers as it’s the most common form of scientific misconduct of late. Authors can use all sorts of tricks, such as flipping, rotating, or cropping parts of the same image to fake findings. Editors are fooled into believing the results being presented are real and will publish the work.
Various publishers are now turning to AI software in an attempt to detect signs of image manipulation during the review process. In most cases, images have been mistakenly duplicated or arranged by scientists who have muddled up their data, but sometimes it’s used for blatant fraud.
But just as publishers begin to get a grip on manual image manipulation, another threat is emerging. Some researchers may be tempted to use generative AI models to create brand-new fake data rather than altering existing photos and scans. In fact, there is evidence to suggest that sham scientists may be doing this already.
What a university learnt from launching a fact-checking initiative in the Philippines
Raksha Kumar for Reuters Institute:
As the Philippines went through its midterm elections in 2019, journalists launched several fact-checking projects to extend their watchdog function and hold candidates to account. One of those projects was Tsek.ph, a fact-checking site launched by media companies, civil society organisations, NGOs and academics at the University of the Philippines.
Two things differentiate Tsek.ph from other fact-checking initiatives: it’s anchored by academics and focuses exclusively on fact-checking elections. The project was launched in time for the 2019 midterm elections and it was active during the 2022 general elections too.
The Gary Lineker tweet scandal shows how the BBC has struggled to adapt to the social media age
Marek Bekerman for Nieman Lab:
During discussions around [Gary] Lineker’s social media conduct, media professionals have frequently referred to the distinction between news and current affairs and other BBC output — and the difference between journalists and other contributors. This distinction, once quite rigid, is increasingly blurred.
[…]
Freelance journalists still had to abide by the BBC’s strict guidelines on impartiality, fairness, and accuracy. But other non-staff contributors had more room for maneuvering, depending on their contracts — the wording of which has always been shrouded in secrecy.
This discretionary nature of contractual arrangements has led to confusion and controversy. Many members of the public do not differentiate between a BBC journalist and a commentator, interviewee, pundit, or studio guest. They are all a BBC voice.
What I read, listen, and watch…
I’m reading Arvind Narayanan’s piece on understanding social media recommendation algorithms for Knight Institute.
I’m listening to NPR’s Code Switch on how horror has evolved for people of colour.
I’m watching the second season of Shadow & Bone.
Reviews, opinion pieces, and other stray links:
How do you decolonise the English language? by Mario Saraceni for Aeon.
Adrian Chiles has a naked lookalike who is making a fortune on OnlyFans by Adrian Chiles for The Guardian.
Why is LinkedIn so cringe? by Trung Phan on SatPost.
Chart of the week
Ipsos conducted a 36-country survey between October and November 2022 to ask 24,471 adults whether AI will go rogue this year. Based on a 36-country survey of 24,471 adults conducted between October and November 2022, an average of 27 per cent of respondents per country considered it likely. India and Malaysia (both 53 per cent) showed the most concern while Hungary (16 per cent) showed the least. The complete data came from Ipsos Global Advisor 2023 Predictions (page 21).
And one more thing
Nudge behaviour: YouGov survey shows two-fifths of UK viewers will wait for a show to end its run before starting to watch. Streaming platforms create a self-fulfilling prophecy. They cancel shows because of low viewership, but viewers won’t start watching new shows for fear of being too invested in something that won’t come back for another season.
That’s a shame because if Star Trek were to be created in this era, it would never have grown into the franchise that it is today. The Original Series, which was ahead of its time when it came out in the ‘60s, started with low ratings and NBC cancelled it after three seasons. However, those three seasons produced 79(!) episodes—an average of 26.3 episodes per season. Today’s typical six- or eight-episode seasons would take three to four seasons to give viewers 26 episodes. So, those 79 Star Trek episodes would be equivalent to ten seasons of The Original Series, if it were to be produced today.
Star Trek was allowed to simmer and eventually became a hit through syndication through the ‘70s and ‘80s. Today, it is one of the largest and most well-known media franchises in the world. Streaming networks bank on instant hits from adaptations of book or video games that already have an existing fanbase such as Game of Thrones and The Last of Us, and original series will hardly likely find the same success with the altering viewer behaviour the industry has created.