🧠Mind expanding 2 Jul 2023: human+AI holobionts🧬🔣embedding all of human belief📚poisonous predictions💸quantum technosocialism🔀💻do octopuses dream?🐙mechanistic interpretability🔬

Encourage people to live inside a helmet

Mind Expanding is my ~fortnightly curation for Memia subscribers of links to deeper dives and bigger thinks and other eclectica that I’ve come across while compiling the weekly newsletter. Thanks for reading!

(A bit later than scheduled hitting “send” this issue… lots going on in the last week…thanks for your patience🙏)

Links in today’s post:

• 🔮🤭New Zealand’s leading futurist (it says here)

• 🧬🔣Human+AI holobionts a shift in biology understanding - humans as meta-organisms including microbiota - extended with AI

• 📚Embedding all of human belief: Project Tenet aims to encode all religious beliefs and more for machine learning

• 💸Poisonous predictions: the inherent philosophy underpinning of McKinsey's ludicrously optimistic AI economic potential projections

• 🧑‍💼AI not killing jobs. Yet. Contrary to fears, AI has not caused significant job losses so far

• 🌍AI to solve global problems: AI's potential in addressing climate change, biodiversity loss, and inequality

• 📉📈The global economy - where to from here? discussion of future global economic scenarios

• 🔀💻Quantum computing for economic modelling: A new breakthrough demonstrates quantum processing for far faster complex economic modelling…could this bring a new age of “technosocialism”?

• 🧠How brains work: Neuroscience and psychology insights:

  • 😶‍🌫️Brain circuits for consciousness: new research has identified the parts of the brain implicated in conscious awareness

  • 🤝The neuroscience of trust how easily hacked we are…

  • 😴Weird dreams an AI-based theory explains why we dream

  • 🐙Do octopuses dream? research shows octopuses display REM-like sleep cycles

• Miscellany

  • 🛍️The end of brands as we know them: The impact of generative AI on marketing

  • 🔬Mechanistic interpretability: Understanding how Large Language Models "think"

  • 🗺️Mapping the future Simon Wardley’s futurist methodology

• 📚 Reading List a few more book recommendations for your virtual shelf

Enjoy!

---

ICYMI

🔮In what could be described as a clear case of the blind leading the blind, this week Aotearoa news outlet BusinessDesk labelled me “New Zealand’s Leading Futurist”(™🤣) when I was interviewed on the latest Business Of Tech podcast: What's in store for Aotearoa, with Kiwi futurist Ben Reid.

(That’s very flattering, I’ll take it.🤭)

The podcast was recorded last Friday just before the NZ Hi-Tech Awards when I sat down with the most excellent Peter Griffin in Ōtautahi and we had a long conversation about emerging technologies and their implications for Aotearoa's future. A really free-flowing and wide-ranging discussion, take a listen. (And I highly recommend the whole Business Of Tech podcast series, Peter and colleague Ben Moore are doing superb work since they started just a couple of months ago.💯)

---

🧬🔣Human+AI holobionts

The Economist ($walled, sorry) explores the concept of “holobionts” — meta-organisms made up of animals, plants AND the microbiota that live on and inside them. This represents a paradigm shift in how we think about biology:

“How many cells are there in a human being?” ...One answer is around 37trn. This is the number, in a typical adult weighing 70kg, that trace their descent from the fertilised egg which brought that human into existence.

Look at it another way, though, and you arrive at a figure roughly twice as large. That adds in the archaean, bacterial, fungal and protist cells which occupy the mouth, gut, skin, lungs and almost every other surface, nook and cranny of the human body. These cells contribute only about 0.3% to a person’s body weight. But being, on the whole, much smaller than “proper” human cells, they are almost equally numerous.

Metagenomics analyses simultaneously the genomes of everything [in a sample of biological matter]...

...The best studied animal holobiont of all is Homo sapiens. Topologically, a human being is a torus—a three-dimensional object with a hole through the middle. The hole in question is the alimentary canal. Nearly the whole surface of this torus is home to microbes, though different parts have different inhabitants. By far the largest numbers of them live in the lower gut.

These gut microbes extend the digestive capabilities of the human holobiont in the same way (though not to the same degree) as happens in termites, by breaking up fibrous plant polymers into smaller molecules that the other 37trn cells can metabolise. But they produce lots of other molecules, too, some of which send signals to the holobiont’s animal cells. Those cells, moreover, often signal back.”

image: maría medem / The Economist

Further to my recent musings about human/AI symbiosis — including that there is an uncanny resemblance between current gut-bacteria/human relationships and future human/AI relationships — my Twitter bio now starts: “Human-AI holobiont”.

(This also implies that the definition of “biont” abstracts beyond just DNA-based genetics to some new kind of gene+memetic bio/digital information substrate…)

---

💭Going deeper on AI

A roundup of deeper (and broader) thinking about the current AI acceleration which doesn’t always make the cut for the weekly scanning emails…

📚Embedding all of human belief

Project Tenet is an open source community initiative to embed all of human belief.

(In machine learning, “embed” refers to the process of converting complex, high-dimensional data into a lower-dimensional, dense representation that captures its key features).

The first visual projections based on a few religions are shown below:

The Alexandria Index is a core component of this initiative which collates together all of the core religious texts and their (machine-readable) embeddings: