Exploring the “AI questions” shaping public opinion

Issue 106 l Eka’s Weekly Roundup (21 Jan 2024)

Jan 21, 2025

London film buffs will know that the London Short Film Festival is taking place this week. I’ve been increasingly curious about what art can teach us about the broader perceptions on AI, as the VC and tech ecosystems can often feel detached from public opinion.

Last night, I attended a screening of The Last Invention, a collection of short films created by or commenting on AI. Here are a few key questions that (I think) the films are posing - and how they relate to impact across ethics, relationships, and work.

1. How do we navigate truth, verification, and trust in an AI age?

We are approaching an inflection point where AI-generated photos and videos will be nearly indistinguishable from real ones. "The Moment of Truth" by David Baksh, explored how photo & video archives are starting to take measures to prevent contamination by AI-generated content and maintain verified content in their physical warehouses.

The rapid evolution of AI tools like Midjourney over the past three years is expected to become even more sophisticated within the next 9-12 months (potential inflection point?).

In the last year, archive workers at Topfoto started noticing something strange.

Even though they had been working with Windrush Generation photos for 50+ years, they started to see completely new colour photos come up on their social media feeds.

These new colour photos had faces which the archivists weren’t able to match onto the original boat passengers…

Source. Ok this is an obvious source but there are many ‘trickier’ ones to look for!

There had been an increasing number of AI-generated images of the Windrush Generation, which had been trained on broader content than just the original archive’s images and were impossible to match against the boat passengers.

Topfoto is now working to secure and link digital & physical photo records in order to maintain trust in their content. This could open doors for even more lucrative license deals with ‘proven true’ archives who will now own verifiable ground truths for major historical events.

The truth matters, especially during a time with declining trust in institutions.

2. What education is needed to make AI adoption more equitable and fair?

There are a few options here:

Verification at the point of source (i.e what Open Origins is doing).
High-quality & verified community notes (i.e. a derivative of Meta is proposing to do - with lots of caveats!)
Longer-term focus on critical thinking and fact checking both at the consumer and creator level.

Those who are able to leverage and critically engage with AI stand to benefit. We need to ensure that all across the ecosystem are able to be in this position, especially if they have traditionally been underserved and/or vulnerable.

We are especially excited about AI tools which help people who are neurodivergent and/or who can use this to alleviate mental load across the household (see a mapping coming soon 👀 ).

3. A tool versus a platform shift. How do creators think about AI?

The directors came on stage after the viewing to pull out themes raised by their films. One made the point that AI was either a tool enhancer (i.e. you still need to know a lot of film-making to utilise it) or a platform shift and totally ‘democratises’ film making (i.e. you don’t need to know a lot of film-making).

I’ve been thinking a lot about this, especially after playing with Suno and making some (very unsuccessful) liquid DnB tracks. I still have quite a long way to go before I can launch a music career…

When thinking about this question and impact, there are a few second-order points which become particularly important:

How will early-career graduate jobs change in response to AI? See our “Key Charts” section for another example or where AI is cutting out early-career work which will have downstream impacts for young people coming into work. A lot of the automation is happening at this end of the jobs market, and people with some CV experience stand to benefit compare to those who are freshly out of university or in apprenticeships.
Does AI continue the ‘fame-skew’ that we’ve seen with the internet, and how does the long-tail economy adapt? There’s this interesting tension between democratisation of technology & the seemingly compounding benefits to existing brands. As technology democratises access, fame anecdotally seems to concentrate even more on particular profiles even if theoretically this should spread more evenly. One film maker said they were most scared of not being seen rather than being stolen form. Powerful!

4. Is AI a mirror to humanity or does it amplify extremes?

Content generation can be used both for extremes and for the average. We’ve spoken about how other technological shifts have impacted mental health, anxiety, and other second-order impacts.

From the Bulletin of the Atomic Scientists - known for their Doomsday clock.

One of my favourite films yesterday was The Oasis I Deserve by Ines Sieulle which explores how humans interact with their Replika personas. I won’t spoil it but there are some pretty outrageous interactions on display…

How does this impact our ‘real’ life relationships?

Side note - AI films are still VERY experimental & art house, so its not for the faint of heart! I’d still definitely recommend going to set a benchmark of how the medium is going to develop - this feels like the real beginning.

5. “These are either the dullest or most exciting AI films created”.

The director of Moment of Truth had this awesome line around this being either the ‘most interesting’ moment for AI films or the ‘dullest moment’ for AI films.

Either way, we expect to see a lot more creative endeavours which help us reflect on AI x creativity x impact outside the normal ecosystem!