#0149: 52+ Things I learned in 2023

Photo by Gary Butterfield on Unsplash

Braingasm

[ED: Happy New Year! This post covers a range of topics that have interested me over the last 12 months. No individual link is particularly technical, but there are some references to technology that might require further explanation if you are not deeply into the topic.]

A couple of years ago (inspired by this), I started trying to find a single thing each week that I could say “this week I learned” about. I managed to do that for all of 2022 and published this post at the beginning of last year (2023).

After trying to learn something each week again in 2023, here are the results of my curiosity. Note: there were a few weeks that had lots happening, so there are a few more than 52 links — at no extra cost — this time around. Enjoy!

1. Here’s a list of ‘extreme questions’ to help trigger new, better ideas during ideation and other creative activities
2. Generative AI tools like ChatGPT could spell the end of the high-school essay, but perhaps that is a good thing
3. Great ideas are hard to find. Drawing makes it a lot easier. And fun. How?
4. Will generative AI kill art or create new opportunities for humans and machines to work together? The weird and wonderful art created when AI and humans unite
5. This week was a very big week in generative AI!:
   - When AI and humans unite, weird and wonderful art emerges
   - 20 Things I’ve Learned in my 20 Years as a Software Engineer
   - The Expanding Dark Forest and Generative AI: Proving you’re a human on a web flooded with generative AI content
   - AGI Is Silicon Valley’s First Native Cryptid
   - How it feels to have your mind hacked by an AI
   - Generative AI and The Future of Work — Unexpected ways generative AI will change how you work forever
   - Who Owns the Generative AI Platform?
6. 61 years before anyone had heard of large language models and ChatGPT, Douglass Engelbart was already thinking about augmenting human intellect. Later in the year, I made the comparison between Doug Englebart’s Mother of All Demos, Steve Jobs’ Introducing the iPhone, and Sam Altman’s recent OpenAI Demo Day. I called it The Granddaughter of the Mother of All Demos
7. This is a great quote from an equally great to-camera thinkpiece on the early-2023 tech layoffs: “AI can only do what human beings have been doing. Only humans can do what no human being has ever done before.”
8. There has been a big uptick in teen mental illness. But it hasn’t happened overnight. It’s been going on for years. And the data shows that the relationship between social media use and mental illness is causative, particularly in young girls:
   - The Teen Mental Illness Epidemic Began Around 2012
   - Social Media is a Major Cause of the Mental Illness Epidemic in Teen Girls. Here’s the Evidence.
9. Matty Benedetto Designed and Prototyped 350 Inventions in 3 Years, many, if not all of them, completely and utterly useless
10. Secret Cyborgs: The Present Disruption in Three Papers. The future is already here, we just need to figure out a few details.
11. The more I read articles like this — and they seem to pop up increasingly frequently — the more I wonder if the “safe level of use” rating for social media for some age groups should be “zero”. It also makes me wonder if we should stop (ban?) tracking behaviour online altogether. I have a simple idea for what we should track instead, but that is a topic for another day. ‘Why the Mental Health of Liberal Girls Sank First and Fastest: Evidence for Lukianoff’s reverse CBT hypothesis’
12. Is GPT-4 starting to exhibit “sparks of general intelligence”? Almost certainly not … quite … but something unexpected is starting to happen
13. Existential risk, AI, and the inevitable turn in human history
14. Generative AI is good at cooperating with people and bad at full automation
15. GPT-4 is starting to exhibit “sparks of general intelligence”
16. Why do ships use “port” and “starboard” instead of “left” and “right?”
17. Animal populations experience average decline of almost 70% since 1970, report reveals
18. Will AI Become the New McKinsey?
19. Where Do Great Ideas Come From?
20. The ‘Beepberry’ is a hackable, portable, DIY handheld that looks a bit like a Blackberry
21. The EU is making an attempt to regulate AI. Will this be a new GDPR moment, or something else entirely?
22. Generative AI is having a big impact on software engineering (amongst other things)
23. In a complete surprise to me, Microsoft PowerPoint is (apparently) Turing Complete. This led me to the realisation that there is such a thing as competitive e-sports Excel. Which, of course, then led me to competitive marble racing, also known as the Marbula One
24. Unison is a new approach to distributed programming
25. Anscombe’s Quartet is an example of four data sets that have nearly identical simple descriptive statistics, yet have very different distributions and appear very different when graphed. Also, the Datasaurus Dozen is an interesting exercise that follows on from Anscombe’s Quartet, which shows that it is possible to generate bivariate data with a given mean, median, and correlation in any shape you like — even a dinosaur. And finally, here is a whole list of statistical methods and concepts with great names that need to be much more well known: Handy statistical lexicon.
26. The AI Canon by A16Z is a great resource for all things AI and gen-AI.
27. ‘Risk Storming’ is a visual and collaborative risk identification technique for lean/agile product and engineering teams
28. Desalination is really difficult
29. Digital minds, if we can call them that, use many different ways to ‘know’
30. This mind-boggling palm flowers underground, and biologists have no idea how it pollinates
31. Wavy walls actually use fewer bricks than a straight wall
32. Artificial General Intelligence — A gentle introduction
33. A lot happened in 1971. And by “a lot”, I mean an awful lot of change kicked off that went on to be globally culturally, politically, and economically significant for the next 40 years or so. Looking at the data, it looks like 2012 might be a similar year.
34. Life on board a British nuclear submarine
35. According to this Microsoft HR veteran, bad managers almost always share these same 4 traits
36. Back in 1962, Doug Englebart wrote (amongst other things) this incredibly prescient paper on ‘Augmenting Human Intellect: A Conceptual Framework’. He was so far ahead of his time.
37. US rejects AI copyright for famous state fair-winning Midjourney art
38. This is a great introductory article on the underlying tech driving generative AI: “Generative AI exists because of the transformer” (FT.com)
39. ChatGPT can now see, hear, and speak
40. A Ulysses pact or Ulysses contract is a freely made decision that is designed and intended to bind oneself in the future
41. Roll Your Own All-Sky, Raspberry Pi Camera: Use Raspberry Pi hardware to capture mesmerising time-lapse images of the heavens
42. The ‘Thumb Calendar’ is a really neat way to represent an entire year in a very compact space, and here is some more on the theme of the ‘compact calendar’
43. Every Software Business Has the Same Playbook — The strange contradiction at the heart of B2B SaaS
44. Embeddings: What they are and why they matter
45. We all know about prompt injection for LLMs. With the advent of multi-modal LLMs (ie text + image), a new form of prompt injection emerges that uses images to escape the LLM’s controls. Here is an example that uses a variation of that attack vector: “Inject My PDF: Prompt Injection for your Resume”
46. ChatGPT Will Disrupt McKinsey Before it Disrupts Google
47. Here’s another article from the same author: The biggest AI headline this week was McKinsey not Open AI. It will be fascinating to see how this plays out in 2024–2025.
48. The largest number representable in 64 bits
49. The surprising connection between after-hours work and decreased productivity
50. Herds of Mysterious ‘Glacier Mice’ Baffle Scientists. Plus, in unrelated news: I was #todaysearsold when I found out that it is trivially easy to combine speakers using the “Multi-Output Device” capabilities of macOS (Sonoma). Here’s a quick guide that shows how to do it. Also unrelated, Role-playing with AI will be a powerful tool for writers and educators.
51. Businesses need to be very careful when they expose gen-AI-powered chatbots to their customers. For example, this guy just bought a 2024 Chevy Tahoe for $1.
52. And a final privacy-reinforcing link to close out the year: How To Turn Off Google’s “Privacy Sandbox” Ad Tracking — and Why You Should

Following on from the 2022 word cloud, here’s one for 2023:

Word Cloud for “Things I Learned in 2023”

That’s it for 2023! I don’t think I’m the only one happy to see 2023 in the rear-view mirror. I trust that 2024 brings you joy and good fortune!

Regards,
M@

[ED: If you’d like to sign up for this content as an email, click here to join the mailing list.]