• StrategicAI
  • Posts
  • Why fake is the new normal, the first AI-discovered drug, meet the digital twins with a soul and why you need to have 'the AI talk' with your kids...

Why fake is the new normal, the first AI-discovered drug, meet the digital twins with a soul and why you need to have 'the AI talk' with your kids...

March 15, 2024

Thought of the week: fake is the new normal

If you missed the story, Kate Middleton posted a Mother’s Day photo that had been digitally altered. So far nothing new, except the fact that this was clearly meant as a proof-of-life image of someone who has mysteriously disappeared from public view. I wish the Princess a speedy recovery, but what struck me and many others was that the picture screams AI-generated. Those kids are just way too happy! ;-)

But seriously, I think that we have reached a post-truth digital moment where nothing digital – from content to videos – can be trusted.

The implication is that new trust and verification mechanisms will be needed to ensure faith in everything from homework to election campaigns.

Why all parents urgently need to have the ‘AI talk’ with their kids

I first had my heart broken when I was 12. I had been attending a succession of boys-only English boarding schools from the age of 7, so girls were the stuff of dreams, largely conceptual and nowhere to be seen, let alone experienced. So when the opportunity to have a US pen pal arose (my memory of exactly how this happened is hazy), I jumped at it. Her name was Christie, and over a few months of monthly correspondence, I got to know her as a smart, kind and curious US pre-teen who attended high school in Little Rock, Arkansas.

Despite written evidence to the contrary, my friends called her my imaginary girlfriend and demanded proof of life, which I duly passed on to Christie in my next letter. A month later and no letter from Christie. I panicked. Had I overstepped the mark? Was Christie horribly disfigured and therefore insulted by my shallow request?

After what seemed like an eternity, my housemaster finally included me in the after-lunch list of boys who had been lucky enough to ‘receive post’. I carefully opened the pink, heart-strewn envelope and began to read from the scented paper that contained the most eagerly awaited news of her Little Rock high school comings and goings. Christie apologised. She had been recovering from a mystery illness so had struggled to write but, critically, had managed to have her picture (which was enclosed) taken.

My heart sank. The picture was clearly cut out from a magazine and, despite my efforts to hide the evidence, word slipped out that I had been catfished, and the teasing only ramped up. To this day, I have no idea if ‘Christie’ ever existed and/or whether this was just some overweight dude sitting in an Idaho trailer park. Given that she went on to ask me if I could return the favour and send a picture of me, ideally in shorts, I suspect the latter.

The point of this rambling story is, of course, the unprecedented risk facing children online. Most teens are digital first and have grown up with no expectation of privacy, and are actually quite clued up. However, when kids can denude their classmates within seconds or be groomed by a real-time deepfake, it’s a case of 'scary new world'. Last week, in what appears to be the first criminal case of its kind, two teenage boys were charged for allegedly creating denuded AI-generated images of their middle school classmates.

This is why, if you have children, you need to have the AI talk.

The first AI-discovered & manufactured drug

This is more like it! Insilico Medicine is a Hong Kong and New York-based biotech start-up that has raised over $400 million to connect biology, chemistry and clinical trial analysis using next-generation AI systems. It has started human clinical trials for an AI-discovered and -manufactured treatment for idiopathic pulmonary fibrosis, a relatively rare but aggressive lung disease. This is exactly what those with rare but ‘unprofitable’ diseases have been desperately holding out for – a reduction in the cost of drug discovery which makes treatments for rare diseases commercially viable.

Soul man: how digital twins are getting a soul

The PR team for Soul Machines did a commendable job of getting global press last week for their digital Marilyn Monroe which launched at the SXSW festival in Texas last week. Marilyn’s avatar is actually pretty underwhelming, but a deeper dive into their website unearthed a digital twin of MMA fighter turned heavyweight boxer Francis Ngannou. Ngannou recently rose to prominence following high-profile fights against Tyson Fury and Anthony Joshua.

Ngannou partnered with Soul Machines to create a digital twin that has been trained on his biographical data and that is powered by ChatGPT. Though it definitely smacks of version 0.1, you can have a fairly fluid conversation with Francis, and you can imagine Sora-like realism within 12 months. We’re on the cusp of creating video-realistic full digital twins that do our digital bidding, attend meetings on our behalf and mentor us through our WFH loneliness. You can try the demo of digital Francis here.

Non-imaginary teddy bears to fight elder loneliness in South Korea

At last week’s Mobile World Congress, South Korean company Hyodol won the innovation prize for its disturbing-looking ChatGPT-enabled companion doll aimed at combatting elder loneliness. The $1,800 AI-enabled doll looks like a normal cloth doll, but it’s actually meant to act as an interactive digital pal for people who are lonely or in long-term care facilities.

Thanks to the large language model stuffed inside the doll, the Hyodol can supposedly hold conversations with its owners, as well as providing health reminders such as when to take medication or eat a meal. The video of desperately lonely geriatrics being fobbed off with a lo-fi teddy bear is a really strange retro mashup, and I couldn’t help thinking of the posters for the ‘teddy bears behaving badly’ horror move Imaginary, which surely has the best tagline of the year: “He’s not imaginary, and he’s not your friend.” Surely the home of Samsung could do better? Or maybe that’s the whole point?

What we’re reading this week

Tools we’re playing with this week

Tavus Beta: Build human-like replicas and text-to-video with APIs.

Beautiful.ai: Putting together PowerPoint presentations is such a waste of scarce human resources, so I got excited by this AI PPT generator. It solves the problem of throwing together a slick PPT in seconds, but it won’t help you communicate effectively until your realise that PowerPoint decks fundamentally suck.

Midjourney.com: The consumer-friendly version of the Discord-based AI image generator. I’m blown away by the images it creates.

That's all for this week. Subscribe for the latest innovations and developments with AI.

So you don't miss a thing, follow our Instagram and X pages for more creative content and insights into our work and what we do.