Issue 124 - February 22nd 2022
Welcome to the many new readers of this newsletter. Each month or so I highlight fascinating stuff from the world of AI, science, psychology and culture which I think you may find interesting. In this edition, I look at how pigs have helped us advance medical science, and how the pandemic has reset, for good or bad, our relationship with work. I also have an update on the Parker solar probe which has officially 'touched the Sun', and how you can use AI to solve Wordle in just one guess. Enjoy!
The robot will see you now
Pigs, it would seem, are at the forefront of medical science, as two big breakthroughs were announced recently, both involving our porcine friends. In the first, the pigs were used to test a new robotic surgeon that performed laparoscopic operations to connect two ends of an intestine in four pigs. So far, robotic surgery has been controlled by human surgeons, who use the machine to augment their capabilities, offering greater precision and stability. As its name suggests, the big difference with the Smart Tissue Autonomous Robot is that it performs the surgery without the need for humans to guide it. As well as being able to cope with the huge variability of living tissue, the system also includes a novel control system that can adjust the surgical plan in real-time, just as a human surgeon would. Early reports suggest the quality of the work is actually better than the human surgeons.
In the second case, the animal has a much more committed role to play: a couple of months ago a pig's heart was transplanted into a human being for the first time. Doctors were able to do this because ten genes in the donor heart had been genetically modified before it was transplanted: three of these genes cause human bodies to reject pig organs, so these were knocked out, whilst six genes were inserted to help control the immune acceptance of the pig heart, and one additional gene was knocked out to stop excessive growth of the pig heart tissue. The patient was ineligible for a human heart transplant so this procedure was his only hope to live. So far, he seems to be doing well. Since there is a huge shortage of transplant organs, this approach could have a huge impact in saving lives - in America alone, 17 people die every day waiting for organs.
The wrong sort of Reset
The pandemic has reset many different aspects and expectations of our lives and, as we (hopefully) start to see the beginning of the end, it is important to understand which of these resets are good and which are potentially damaging, especially if we don't realise how damaging they are. One of the relationships that has definitely changed for many is their one with work - some have not been able to work much or at all during the pandemic, whilst others are having to work much more (ask anyone who works for themselves). What everyone can agree on though is that it has been hard. As we do start to ease restrictions, many people are reassessing their relationship with work, and how much of their identity is tied up in their job. This great article from The Atlantic asks that most fundamental of questions: Who would you be if work was no longer the axis of your life? For some people, having their work define their whole identity is ok, and that's fine. But for many of us, finding the time to think about the things that make us happy is hard enough, let alone actually doing those things. And that feels wrong. I enjoy my work - I have autonomy, I learn new stuff every day, I get to talk about interesting things to interesting people - and I have hobbies and interests, but the restrictive and habitual nature of the pandemic has taken away the ability for me to get the balance right between the two. I could go out to a gallery, but it's easier to stay indoors. I could take a day off, but it's easier to carry on working. As the article says, "When the haze of burnout begins to clear, fight the urge to feel productive and channel that into beginning to explore your own pleasures". This is really good advice, and it will certainly help to reassess, and potentially reset, the relationship between our identity and work. It's just that it will probably take a little time to get there.
Set the Controls for the Heart of the Sun*
It was way back in September 2018 (issue 89 of That Space Cadet Glow) that I reported the successful launch of NASA's Parker probe which was heading to the Sun to investigate our star from close range. And, just before Christmas 2021, that probe flew through the Sun’s upper atmosphere (the corona) and sampled particles and magnetic fields there. It managed to get within 6.5 million miles (or 15 solar radii) of the Sun's 'surface'. Parker will continue to spiral closer to the Sun, eventually reaching as close as 8.8 solar radii, or 3.8 million miles. The data collected will help scientists understand the nature of solar winds (which hugely impact the Earth) and also try to answer the mysteries of what causes 'switchbacks' (S-shaped kinks) in the solar wind, and why the corona is much hotter than the surface of the Sun. This 5-minute video from NASA gives a great summary of the findings to date.
* Readers can win a dose of smugness if they can spot the connection between this headline and the newsletter's title.
Cracking Wordle
Most of the English-speaking world (myself included) seem to be obsessed with the daily word puzzle Wordle. It would therefore be remiss of me (I make no excuses for the level of geekiness expressed) not to share with you a couple of examples of where people have used deep analytics to try and solve Wordle in the most optimum way. Firstly is 3Blue1Brown, whose always entertaining videos explore maths and statistics using animation. He uses Wordle as an excuse to explain Information Theory i.e. what word provides the most 'information'. (There is also a supplementary video that goes into more detail about the ideal starting word). A very different approach is taken by Ben Hamner, who has developed a little application that uses the shares on Twitter of people's results (which just show the coloured squares without the letters) to calculate the correct answer in the first guess. You can get the code yourself from his Kaggle page. Obviously, both these techniques take all of the fun out of the game, so are not recommended, but they do demonstrate how information is inherently stored in the data, and how it can 'leak out' even when seemingly innocuous information is shared by many people.
Just because you can, doesn't mean you should
I recently gave a webinar looking at how AI Ethics can be applied practically in the real world (you can watch the whole thing by registering at this link). In the webinar, I gave some examples of where AI had been used inappropriately, and, in some cases, at the expense of those it was actually trying to help. Well, just this weekend, another prime example of this jumped out at me from the pages of the newspaper. Shout is a charity crisis text line for people needing urgent mental health support, and as you would expect, is confidential. But they have been sharing anonymised versions of their data with researchers who are using AI to better understand the types of people that are using the service, as well creating tools to predict behaviour, including, for instance, suicidal thoughts. The article talks about whether the charity had properly informed their users or not, but there are bigger questions than this: at the heart is simply understanding what is the right thing to do. The charity seems to have forgotten why they exist in the first place and have been tempted into using AI to, ostensibly, help improve the service. All they have managed to achieve, though, is to completely erode the trust that was fundamental to the service, and therefore cause more harm than good. This is a really important lesson to learn for anyone (like me) who works in AI: just because you have the data, and using AI may help a bit, really doesn't mean you should use it. That should be Question 1 on any evaluation of using AI.
Bonobo - Fragments
There is something undefinable about Si Green's music. Reviewers compare it to chill-out masters such as Air, Röyksopp, Cinematic Orchestra or Caribou, and there are elements of all of these in the tracks on Fragments, his latest LP under his pseudonym of Bonobo. There is a greater variety across the album than any of those other artists achieve, but it still retains a distinctive feel, and this, I think, is what makes it so listenable. It's a record you can put on most times of day and whatever mood you happen to be in it will sound good. Highlights such as Rosewood (which has clear influences of Bicep) are matched by the collaborations with artists such as Jamila Woods on Tides and Jordan Rakei on Shadows. You can listen to the whole LP on Spotify or Apple Music and buy it on Bandcamp.
Greenhouse Intelligence Ltd thegreenhouse.ai
Andrew Burgess is the founder of Greenhouse Intelligence, a strategic AI advisory firm.
You are receiving this newsletter because you subscribed and/or attended one of our events.