SXSW 2024: Amy Webb and the "Transformation Generation" vs. IO

And it took another fifteen minutes due to a technical problem with her presentation (Amy Webb would admit to being guilty of the "I-don't-send-my-slides-in- advance-because-I-want-to-update-them-until-the-last-minute" syndrome), before the audience could discover the "not-trending trends" (trends not trendy). It details 3 main ones, taken from the 1000 pages of the Future Today Institute report, which brings together both new technologies and industries: Artificial Intelligence, a technology for general use in the same way as electricity, the steam engine or the Internet, the connected ecosystem of objects and biotechnology.

FUD (Fear Uncertainty Doubt) in the face of the technological supercycle

And as if to reassure her listeners in search of answers, Amy Webb declares that we are indeed living in "the most complex operating environment" she has seen in 20 years, a new "Technological Supercycle". And the problem of our time is that leaders, overwhelmed by the speed of change in the world around us, filled with FUD (Fear Uncertainty Doubt), make decisions based on a combination of simple fear and FOMO (Fear of Missing Out) without even having started their digital transformation.

We are all in this together, gathered in a new generation: the Transformation Generation, which is being forced to transform at a much faster rate than it can comprehend what is happening to it. Even if there are still elements of stability. Last year, Amy Webb tested LLMs on their tendency to bias (which only reflects real-world dysfunctions) by asking them to draw the image of a CEO. You can't get anything out of it but a white man with grizzled temples.

She repeats the exercise a year later and even twisting the prompt by adding that it is a CEO of a factory that manufactures women's hygiene products, the result remains the same:

Illustration of a CEO of a tampon factory, as seen by ChatGPT4

One day after Women's Day, there's definitely progress to be made, whether it's Midjourney, DALL-E, or Anthropic. The first AI trend is therefore logically responsibility. Ethical teams are still not considered stakeholders, and ridiculously aggressive Google Gemini-style reinforcement learning doesn't solve the basic problem either.

The second trend mentioned around AI: the "Concept to Concrete". After text2text, text2image, text2code, and even text2flirt, the trend ranges from literal instructions for literal answers to concepts for assisted thinking (Pica, Sora are some examples). No need for prompt engineers, we will soon be able to think about what we want AI to do. All this in the face of the third trend: Insecure AI. Gone are the pioneering days when companies leaked scientific papers related to their work on generative artificial intelligence, as the financial stakes were far too high.

Presentation of the Future Today Institute

Contrary to what is cherished in Europe, American culture does not look favourably on Open Source models like Llama 2 (we remember the episode Llama 2 Uncensored). Amy Webb still evokes the wind capable of clearing the sky in the Rhône Valley, talking about Mistral, while being surprised that the French start-up has published its model in the unusual form of torrent (peer to peer format), going against any management of intellectual property rights and security and control, proof for the director of the Future Today Institute that insecure models would get crazier and crazier.

And for now, AI doesn't need a license, it doesn't go to jail in case it hurts someone.

In the meantime, AI needs data to power itself, which brings us to the next trend: LAMs, Large Action Models. In the connected ecosystem of Things, we are soon reaching the end of the available data, LLMs need new data. The solution: sensors everywhere, wearable connected objects, supposed to make us interact in innovative ways, but secretly capturing all our emotions to feed large models - no longer of language - but of action, and to predict our desires. Amy Webb talks about a Cambrian explosion of connected devices, with a few that will seem weird, but others (like PockEngine, or Rabbit R1 presented at CES this year) that will be able to process data that is not text, and that will be our new personal portable neural network, a kind of second brain in the pocket (a smartphone to the power of 100), and who will learn from our everyday actions.

A fifth trend, which goes in the same direction: Face Computers everywhere. We're talking about the Apple Vision Pro, a computer that we attach to our faces, and which, according to Amy Webb, in 18 iterations (what it took for the iPhone to become mainstream) could conquer the world of objects connected to our emotions and brains that power the space Internet. So the next step is the "Battle for Face Supremacy". The human becomes the next interface, with undesirable future scenarios: What happens when my grand action model hallucinates? What if connected objects led to a Chinese-style social credit system?

The supermarket of the future imagined by Amy Webb

We could use our attention span for rewards, for example at the supermarket. The digital divide will then become even more economical, between those who will be able to pay for an Adblocker in their headset, and those who will have to wait for the passage of advertising before being able to make their purchases so as not to fall victim to dynamic pricing (no attention, no coupon).

Or we can even use our life sequences to predict death. Welcome to the Large Death Model, the holy grail of any insurer.

For the media industry, the direct impacts (in light blue, the furthest are in red, we would have reversed these color codes) are on the side of: AI, generative AI, AR/VR/XR, metaverse (it's still there) and Web3

AI and news: Focus on value, not tools

When it comes to technologies implemented in newsrooms, according to the Tech Trends Report, the focus should be on value creation rather than the tools themselves. When start-ups launch new AI applications on a daily basis, their lasting impact on information remains uncertain. It is therefore crucial to assess how these technologies are changing value creation, especially in news, and to understand the implications of integrating these technological tools into the editorial process and the media value chain.

Thanks to artificial intelligence, readers can access an unprecedented amount of information. The ability to explore and synthesize information effectively can establish a new intermediary between publishers and their audiences. Media outlets are wondering if they can still capture their audience directly or if the public doesn't prefer to get their information via automatic summaries in search, conversational interfaces or new platforms that we don't even think about yet. In any case, the formats in which the public consumes information will be increasingly varied and fluid.

Blacklight by The Markup

In this new context, media outlets need to hire employees with technical skills similar to those sought after in tech companies. When journalists master technical skills (in addition to their editorial acumen), it leads to impressive achievements. The Markup spent 18 months developing Blacklight, a tool for analyzing trackers on websites, while The Wall Street Journal built a network of bots to deconstruct and describe how TikTok's algorithm works.

Generative artificial intelligence tools are reducing barriers to the spread of fake news on a large scale, making the search for effective solutions to disinformation and misinformation increasingly urgent. This urgency is exacerbated by AI's propensity to assert inaccuracies with confidence. Google has integrated a feature to "verify" Bard's (now Gemini) answers via a color-coded fact-checking interface. Bing's chat experience can provide footnotes to back up its claims. Solutions of this kind will be crucial to increase consumer confidence in the results provided by AI and to prevent AI "hallucinations" from hijacking public debate.

Sensory Journalism

With the advent of immersive technologies in the mainstream, journalists are acquiring a new ability: that of telling stories that directly engage the senses of listeners. This new narrative horizon, which forges deep emotional connections between the protagonists of the stories and the audience, comes with new ethical challenges that journalists will have to understand and raise.

News scenario 2027 (rather 2025, editor's note):

What if newsrooms were replaced by an AI that could summarize the world? The rise of generative artificial intelligence technologies is radically transforming the way information is consumed, with advanced language models that offer personalized news summaries in real-time. Systems like iOS and Android now incorporate news briefings based on users' personal interactions and international news, creating painstaking summaries that encompass a variety of sources, from social media to news sites and far beyond. This personalization pushes user engagement, making devices a central pivot in the consumer's understanding of the world.

This innovation poses major challenges for traditional media. Few newspapers manage to negotiate profitable licensing deals with tech giants, while the majority struggle for survival, unable to prove the impact of their work in an ecosystem where platform transparency is lacking. Direct interaction between consumers and traditional media is declining, raising questions about the future of quality information and the responsibility of platforms in disseminating news.

The latest trend presented by Amy Webb is Biotechnology. As Moores's law seems to be failing, DNA and RNA will come as a knight in shining armor. Technologies like Groq (not to be confused with Elon Musk's Grok) could allow ChatGPT to run 14 times faster. Generative biology (DeepMind GNoME) leads us to Organoid Intelligence (IO) which uses pieces of organic tissue, as in an experiment at Johns Hopkins University. We will be able to make biocomputers out of human brain cells, and grow computers rather than build them, even from our own cells. How about taking a vacation?

Organs-on-a-chip are small devices that contain tiny pieces of human tissue inside, and they are specifically designed to keep tissues functional as they would be in the human body. Image Credit: Penn Medicine News

It seems that it is not the technology that is the problem, but the way people use it. Can we trust the tech messiahs who may soon be conducting biotech tests in poor economic areas? Amy Webb calls on governments to set up "Transition Departments" ("it doesn't matter how old they are"), and for companies to remember their ecosystem of values. Its presentation made it possible to put new words on trends and realities already known, which nevertheless remain quite pixelated. Maybe we should ask ChatGPT how best to follow Amy Webb's advice to "fight for our future." In any case, we have to hurry, the Future has already arrived today.

Images generated by The Future Today Institute, with Midjourney

To access the thousand pages of the report (from page 572, the News section)

This article is written by Kati Bremme, Head of Innovation at France TV.