Future Friday: AI Beats You at Mafia (and Writes the Press Release to Go With It)
Plus: The Deep Research Wars heat up, why code hallucinations might be our smallest problem, and so much more.
Dear Friend,
It’s Future Friday (again) – and the world clearly didn’t stop spinning this week. While AI keeps our newsfeeds full (and us busy), many other technologies are advancing at equally breathtaking speeds. From synthetic biology to the energy transformation – yet, this ought to act as a good reminder that not everything that shines is gold; some hyped technologies, such as eVOTL (electric vertical take-off and landing aircraft), are pretty much going nowhere. I guess we all have to wait a little longer for our jettison-fueled future…
And now this…
(*) The eVOTL news is courtesy of our wonderful radical Innovators & Disruptors community.
Headlines from the Future
The present moment was once the unimaginable future. — Steward Brand
—//—
The Widespread Adoption of Large Language Model-Assisted Writing Across Society ↗
A recent paper from Stanford University shows how prevalent the use of LLMs such as ChatGPT is in the creation of text. By late last year (2024), a whopping 18% of financial consumer complaint text and 24% of corporate press releases (not all that surprisingly – though it does make you wonder why you would pay a PR agency to do this anymore) appeared to be LLM-assisted. Interestingly, only about 10% of job applications appeared to have been written (or enhanced) by AI.
Our study shows the emergence of a new reality in which firms, consumers and even international organizations substantially rely on generative AI for communications.
—//—
The Differences Between Deep Research, Deep Research and Deep Research ↗
Confused about the difference (if any) between Deep Research, Deep Research, and Deep Research? That is, the difference between OpenAI’s, Google’s, Perplexity’s, and the many other products all calling themselves “Deep Research”?
It seems Deep Research is the Retrieval-Augmented Generation (RAG) of 2025—everything is being rebranded and marketed as “Deep Research” without a clear definition of what it actually entails. Does this sound familiar? It echoes the hype around RAG in 2023, agents, and agentic RAG in months past.
Read Han Lee’s explainer for an in-depth comparison.
—//—
LLMs Are Very Good at Playing Mafia ↗
Mafia is a social deduction party game where players are secretly assigned roles as either innocent “villagers” or murderous “mafia” members.
During “night” phases, mafia members secretly choose villagers to eliminate. During “day” phases, all surviving players debate and vote to execute someone they suspect is mafia. The game continues in these day/night cycles until either all mafia members are eliminated (villagers win) or the mafia outnumbers the villagers (mafia wins).
The fun comes from deception, strategic accusations, and trying to identify who’s lying based on behavior and arguments.
It turns out that AI is rather good at playing Mafia, which makes you wonder what this all means for humanity. 🤷🏼
Link to LLM Mafia Leaderboard.
—//—
The Tides of Change Are Unstoppable... ↗

—//—
Hallucinations in Code Are the Least Dangerous Form of LLM Mistakes ↗
The ever-brilliant Simon Willison on the challenges we face with LLMs and their use as coding assistants (so much for “vibe coding” – which is a truly idiotic concept, by the way…):
Hallucinations in code are the least harmful hallucinations you can encounter from a model.
The real risk from using LLMs for code is that they’ll make mistakes that aren’t instantly caught by the language compiler or interpreter. And these happen all the time!
[…] Compare this to hallucinations in regular prose, where you need a critical eye, strong intuitions and well developed fact checking skills to avoid sharing information that’s incorrect and directly harmful to your reputation.
Read the whole thing; it has some good insights for non-coders as well.
What We Are Reading
🤝 Using AI to Crack Management Challenges Think of AI as your management co-pilot. With the right framework and a healthy dose of skepticism, you can tackle management problems in entirely new ways. @Jane
💰 The Secret of Success: A Study of 50 Million Businesses Reveals Why Some Companies Thrive — and Others Fail The most significant factor contributing to a business's success is the amount of financing it is able to raise before launching, with venture capital investment having the most positive impact. @Mafe
🎓 How I Use LLMs If you are looking for a great expert-led, 2-hour crash course on LLM use cases with loads of practical examples, this new video from Andrej Karpathy is for you. @Jeffrey
🛸 Should We Be Moving Data Centers to Space? Isn't it interesting to see rather concrete utilizations of space in novel forms? While it's always been looming to better leverage the potential of space, much has been arbitrary or simply ridiculous. Mainly driven by physical limitations and ecological concerns, but now mixed with considerations of additional security for the infrastructure, data centers above the clouds are now being prototyped. @Julian
🌐 How Open-Access Networks Boost Your Internet Provider Options An open-access fiber network is a type of internet infrastructure where a single company installs and operates the network, then sells access to various internet service providers (ISPs). An example is Gigapower, a joint venture between AT&T and BlackRock, which is expanding fiber into previously overlooked communities. @Pedro
🎭 These Are the Storytelling Lessons I Learned From Steve Jobs One more thing: Creator of the iPod and Nest, Tony Fadell, recalls the lessons learned from master storyteller Steve Jobs. From persistence to the "virus of doubt" – Steve truly excelled at the art and craft of, well, telling a good story. @Pascal
Some Fun Stuff
🎶 You already know that vinyl is back. But did you know that there is a whole segment of the fan base that cuts their own vinyl records?
💦 Gorgeous collection of, yes, fire hydrants. Sortable by color, shape, and many other criteria.
🛫 Mesmerizing: Listen to live air traffic control radio chatter mixed with ambient music.