It seems to me that when we fail, it’s because our imagination fails us. I believe this to be true of both life and literature.”* —Claire Keegan [1]
Dear Reader
Welcome to 2025, a year on the edge.
Edge of what? Well, that’s the whole adventure, isn’t it? My advice to myself—and you’re welcome to borrow it—don’t be tempted to wait for the year to reveal itself. Dive in and explore. Start making it into something.
Palette Cleanser for the Mind Let’s begin with something soothing yet intriguing.
This is the Idirisi Ensemble, a choir that recreates ancient chants and songs dating back to Roman times. While they’ve only released a single track so far, their TikTok and Instagram accounts hint at a treasure trove waiting to be explored.
I am - sort of – not doing predictions. See the footer for an experiment in getting AI to predict AI trends.
We’re going to use a few military metaphors now. We are living through the early days of the AI revolution. The first thing to remember is that we don’t really know what’s happening—and neither does anyone else. I find it useful to invoke the idea of the fog of revolution.
The "fog of revolution" describes the profound uncertainty and disorientation during periods of rapid technological or social change. Like the "fog of war," it refers to the difficulty in seeing clearly what’s happening while you're in the midst of transformative change.
Key aspects of the fog of revolution include:
Difficulty distinguishing between temporary disruptions and permanent shifts.
Challenges of making decisions when traditional frameworks no longer apply.
Inability to assess the impact of changes as they’re occurring.
Tendency to either over- or under-estimate the significance of new developments.
We can respond locally to things we see and make decisions accordingly. We can form a mental picture of the whole context based on communications. But our impression of the whole is at best imperfect and at worst seriously misleading.
This concept is particularly relevant today as we navigate simultaneous revolutions in AI, climate technology, and social structures. The fog makes it challenging for leaders and organisations to chart clear paths forward.
But they still have to keep moving.
Learning From History As It Happens
To navigate today’s complexities, we must become students of the present moment—mindful of both the immediacy of now and the lessons from those who studied their own present moments before us.
This practice resembles the work of Contemporary Historians: those who observe, gather evidence, and look closer at the details, seeking insights rather than stories. Unlike journalism, which claims to write the “first draft of history,” this approach prizes the patterns and meanings beneath the surface rather than the narrative alone.
When I studied at Groningen University in the early 90s, I was tasked with researching the Russian diaspora in Moldova after the fall of communism. What initially seemed like a dull brief became a fascinating lens into larger movements—revealing connections to Moscow, Berlin, and Brussels. It was fascinating, and now newly, grimly relevant again.
History is happening now. To learn from it, we must pay close attention. I recommend you take notes.
Remember December?
December can feel like forever ago as we shake off the hibernatory instincts of Christmas. But it was only a few weeks ago, and as we noted in the This Year is History Edition, the Big Tech players let slip a small avalanche of new tools and features.
Here are the ones I’m exploring now:
Google is back in the game: After a year of embarrassing set-backs, Google launched useful tools and promising upgrades to its ChatGPT rival, Gemini.
ChatGPT’s app follows you around: It becomes digital sidekick and even writing to apps like Notion (Mac version only) and Apple Notes (iOS-only).
Your phone can see: OpenAI’s iPhone app can have a voice conversation with you in real time about what it it can see through the camera.
Canvas on ChatGPT: Creating a kind of document for text, data or coding to collaborate in with ChatGPT is incredibly useful.
ChatGPT-o1: The “reasoning” model brings PhD level thinking to problems - but takes a bit of AI literacy to use effectively - there’s a $200/month even higher powered version – but you need some success with o1 to be even thinking about that. The o3 model will be coming later this year, which is, er, even better.
Gemini Deep Research: A potential Perplexity killer, or a complement? At the moment this is only available on personal Google accounts Gemini Advanced subscriptions. We’ve tested it out and it is seriously impressive. But takes its sweet time. (See our Edition Extras Notion page for more on this.)
First Deep Research makes a plan for your research project.
Which you can edit or amend before it starts work.
And then it says it will get back to you. This one took about 15 minutes.
DeepSeek: Meanwhile, a Chinese competitor to OpenAI has appeared: DeepSeek. It appears to be “reasoning” like ChatGPT-o1 and is very transparent about it, you can watch it have a conversation with itself as it works the query through. See the Notion Extras for more on these results.
DeepSeek is very fast, and has web search option ideal for research...
A full listing of December announcements and analysis is going out to Brilliant Noise clients this week. If you’d like a copy, reply to this email, comment or DM me.
It’s going to be a year of change, shocks, and surprises. I’m looking forward to writing about some of it and sharing it with you.
If you enjoyed this week’s edition, please spare a click to turn the 🤍 into a ❤️ below.
Antony
Footnote:
[1] Claire Keegan wrote one of Antonym’s fiction best reads of 2024, Antarctica, and the novel Small Things Like These, which she adapted into a film likely to earn Oscar nominations.
Extras!
To stop this newsletter getting too long, I’ve put the screenshots and analysis in a Notion page here.
One thing that I will just leave here though is…
AI’s Predictions About AI
A summary of the Google Gemini Deep Research predictions for 2025 that we created (a full report is available on Notion or the whole Google Doc here).
