Microsoft claims their new medical tool is “four times more successful than human doctors at diagnosing complex ailments”. Unsurprisingly, what they meant by “diagnosing a disease” was the thinking-hard part, not the inputs part:

To test its capabilities, “MAI-DxO” was fed 304 studies from the New England Journal of Medicine (NEJM) that describe how some of the most complicated cases were solved by doctors. 

This allowed researchers to test if the programme could figure out the correct diagnosis and relay its decision-making process, using a new technique called “chain of debate”, which makes AI reasoning models give a step-by-step account of how they solve problems.

If and when deployed, how likely is it that these algorithms will get a query comparable to a New England Journal of Medicine case study? Most doctors don’t reach those levels of perception and synthesis, let alone the general public.

📚 Finished reading: A Thousand Brains by Jeff Hawkins, having no idea how it ended up in my Kindle library. I am glad to have opened it, as I now have some semblance of a framework for how this thing we call intelligence might work. Note that the newest developments in neuroscience are just a starting point, as most of the book deals with their implications for AI and the future of humanity. If that sounds like overreach, know that by the end of the book it is. Still, these wafer-thin speculations don’t detract from the book’s meatier parts.

Confirmation bias alert: the framework repeats almost word for word the thought I had a while back — and more recently — about AI, that true general intelligence needs to be able to interact with its environment. So I may be blind to some obvious deficiencies in the argument. But then again, great minds, etc.

I will have more to write about this soon (ha!), but until the stars align for an extended writing session here is a good opinion piece from FT’s John Thornhill about why LLMs may not be all that great for lay people dabbling in, for example, medicine:

When the test scenarios were entered directly into the AI models, the chatbots correctly identified the conditions in 94.9 per cent of cases. However, the participants did far worse: they provided incomplete information and the chatbots often misinterpreted their prompts, resulting in the success rate dropping to just 34.5 per cent. The technological capabilities of these models did not change but the human inputs did, leading to very different outputs.

The emphasis is mine, because it is a neat summarization of what I wrote 2 years ago. Humans are unique not because of what’s inside our heads but because of how we interact with the environment. There will be no artificial general intelligence until that problem is solved.

An excellent blog post that is not a rant: Single-function devices in the world of the everything machine, by Christopher Butler.

Limitations expand our experience by engaging our imagination. Unlimited options arrest our imagination by capturing us in the experience of choice. One, I firmly believe, is necessary for creativity, while the other is its opiate. Generally speaking, we don’t need more features. We need more focus.

Indeed.

After four months of waiting, I’ve received a Daylight tablet yesterday. The good: the e-ink display is better than I expected and writing with the pen is as smooth as can be. The bad: at the end of the day it’s Android and needs a Google account for everything. Also, it’s on the heavy side.

Not a tech or finance expert, but whatever Jony Ive ends up making has the potential to destroy Apple.

When Ive left the company I wondered if he was tired of work in general, or just of Tim Cook’s profiteering. Guess I got my answer.

Today, I learned about The Chandler Project, a doomed attempt to build a next-generation “personal information manager”, and it is wonderful. Just look at this beauty! Sadly, the project went bust more than a decade ago — the last update was in 2009 — but many thanks to whomever is paying to keep the lights on.

The initial development and decline of Chandler was described in the book Dreaming in Code by Scott Rosenberg which, yes, is now on the pile. (↬Thinking With Tinderbox)

Speaking of Gioia, I love his work, his most recent article is just wonderful, and I absolutely share his view on techno optimism… but blogging from Substack makes things a bit awkward.

📚 Thinking With Tinderbox continues to pay dividends, even though I am not learning anything about the app’s mechanics. One of the footnotes led me to About This Particular Outliner and its parent, ATP Macintosh and now I am thinking about the greatness of pre-2016 Internet. Quite the rabbit hole.

It took me four years to drop my Apple Watch habit, but drop it I did. Goodbye, constant notifications. Farewell, nudges to breathe and to stand up and to convert my walk to the grocery store into “an activity”. I will hardly miss you, phantom vibrations and the pale white band around my wrist. You were good for heart rate and pace tracking, and for that you can still sit in the drawer, awaiting my next run.

The de-watchification of my everyday life began a few months ago when I forgot to take it off the charger after leaving it there for the night. This in itself was an aberration as I tended to keep it on at bedtime for sleep tracking ⊕ The number of times I checked the results of this tracking is, of course, zero. This is also how many valuable insights on my sleep patterns I received from Apple’s Fitness app. and only charge it for a half-hour in the morning. I failed to notice a change on that first watchless day, or on subsequent days. A $30 Casio — itself an indulgence since similar performance could be had for under $10 — gave time just as well and did not require charging. With luck, I may eventually get to cleaning and repairing a slightly more substantial timepiece I got some 20 years ago, victim of an inept shopping mall jeweler trying to replace its battery.

This is not the only way I tried to introduce more friction into my life — see the iPhone dumb-down of a few months ago. Kyla Scanlon’s latest article, ⊕ If you haven’t yet checked out Kyla’s blog, please do so now. It is for economy and finance what Ruxandra Teslo’s blog is for biology. The Most Valuable Commodity in the World is Friction, outlines the reasons why one should think about more friction better than I ever could. There is a clear distinction between the frictionless digital and the friction-full physical world, only the frictionlessness of the digital realm is largely an illusion, a sleigh of hand, for:

… we have a world where friction gets automated out of experiences, aestheticized in curated lifestyles, and dumped onto underfunded infrastructure and overworked labor. The effort doesn’t disappear; it just moves.

It was easy enough to nod my head in agreement for I thought of this every time I ordered my groceries to be delivered. But in Apple Watch’s nudge economy the underfunded infrastructure was my calendar and the overworked labor was me.