Making (Some) Sense of Artificial Intelligence

A collection of some of the best links from around the web, manually curated.

Great links like these, every Friday in your inbox

Artificial intelligence is transforming our world — it is on all of us to make sure that it goes well

Max Roser from Our World in Data makes the case for everyone of us to become more interested in how artificial intelligence is transforming out world: «If you and the wider public do not get informed and engaged, then we leave it to a few entrepreneurs and engineers to decide how this technology will transform our world.» He explains his reasoning along three key questions: Why is it hard to take the prospect of a world transformed by AI seriously? How can we imagine such a world? And what is at stake as this technology becomes more powerful?

From Weekly Filet #433, in March 2023.

ChatGPT Is a Blurry JPEG of the Web

If you’re annoyed by the current hype around large language models (the kind of artificial intelligence that can write and chat with humans), or if you still haven’t quite understood how it works, this article is for you. Science fiction writer Ted Chiang offers a lens through which to understand what’s problematic about these models. And he offers a helpful proxy: We will know a large language model has become reliably good in quality when the output it generates will in return be used as training material for new models (which currently isn’t the case).

From Weekly Filet #429, in February 2023.

AI Cookbook — Techniques to improve reliability

The current hype phase of artificial intelligence makes it easy to find confirmation for your preconceptions. It’s easy to find examples that make it look extremely capable. It’s just as easy to find examples where it looks embarrassingly clueless. Where it gets really interesting is in between: the techniques humans can use to elicit more accurate answers from the AI when it fails on its own (for example, by telling it to «think step by step» or, funny enough, «don’t make stuff up»).

From Weekly Filet #426, in January 2023.

Generrated

This is extremely fascinating and a great way to get a glimpse into the creativity of artificial intelligence. Five themes — a representation of anxiety, an astronaut, the discovery of gravity, a horse, someone gazing at Mount Everest — turned into images by artificial intelligence in a myriad of ways: in the style of famous painters and photographers, as company logos and app icons, as movie posters or New York Times front pages from various decades, as Disney characters and bronze statues, and many, many more.

From Weekly Filet #424, in January 2023.

Generative AI: autocomplete for everything

Having played around with GPTChat, and having seen dozens of examples of how powerful this latest iteration of artificial intelligence has become, I can’t shake the feeling that we’re witnessing the beginning of a new era in human-computer-interaction. This joint analysis by an economist and an AI researcher provides a good high-level perspective. Key quote: «Applying the idea of comparative advantage at the level of tasks instead of jobs, we can see that there will always be something for humans to do, even if AI would do those things better.» Strikes me as both fairly depressing, and overly optimistic.

From Weekly Filet #421, in December 2022.

I trained an AI chatbot on my childhood journal entries…

Imagine you could chat with your childhood self. What would you like to ask? What could you learn about your current self? How would it feel? This is the story of coder/artist/scientist Michelle Huang who fed an artificial intelligence her childhood diaries and taught it to emulate her «inner child». With all these AI models rapidly progressing, reality is catching up with science fiction faster than ever. I’m equal parts intrigued and scared.

From Weekly Filet #420, in December 2022.

40,000 Recipes for Murder

Spine-chilling, to say the least. This is the story of two scientists who developed artificial intelligence technology to discover medicines for rare diseases. Basically, the machine tests any and all combinations of molecules imaginable, to find ones that could work as a drug against certain diseases. One day, they realise they could — with just a few adjustments — make the machine do the opposite: Find the most lethal combinations of molecules. It found thousands, some orders of magnitude worse than what’s previously been known. What now?

From Weekly Filet #409, in September 2022.

We Need to Talk About How Good A.I. Is Getting

Smart people disagree over whether artificial intelligence is still mostly overhyped or on the cusp of surpassing human capabilities (and whether that’s good or bad). But there is no denying that a lot of noteworthy progress has happened lately and that we should pay attention. This piece gives a good overview of where things stand — one of the main things to consider, in my opinion, is that we humans are just very bad at grasping exponential change. Things can move slowly for a long time and then change very quickly very fast. (Gift link that gives you free access without a subscription)

From Weekly Filet #406, in August 2022.

The “most important century” blog post series

Let’s zoom out for a bit. Could this century be the most important for humanity ever — of all that ever were and ever will be? This long and thoughtful series of posts makes that argument. In a nutshell:

  • Artificial intelligence will lead to an unimaginable explosion of scientific and technological advancement.
  • When that happens, humans as we know them will no longer be the main force in world events. Instead, it will be digital people or misaligned AI.
  • This could happen as early as this century. Therefore, now is our last chance to shape how that transition happens, making this the most important century for humanity.

To be honest: I went into this series like 🙄 and now I’m like 🤯. These are not simply Sci-Fi musings, but conclusions drawn from research (speculative in nature, sure, but well argued).

You can choose to read a summary or the whole series, you can listen to the series read by the author, or hear the author explain his theory in an interview with Ezra Klein — whatever works for you. Trust me, diving into this is worth your time.

From Weekly Filet #392, in April 2022.
Thousands of curious minds trust the Weekly Filet for food for thought. Every Friday, carefully curated recommendations on what to read, watch and listen to. Helping you make sense of the world and yourself.

Undecided? Learn more | Peek inside