There’s a typically brilliant piece by Liam Shaw on the LRB blog right now about the recent use of an AI tool to assist in the discovery of an antibiotic: abaucin.

There is much important detail in Shaw’s blog which was missing from most of the media coverage on this topic. Most crucially from a health perspective, this antibiotic is likely to be useful only in topical applications (onto the skin) whereas the majority of harm from the single species the antibiotic treats—Acinetobacter baumannii—is from sepsis. It is a significant discovery, but mostly in the sense of being a staging post on the long road of development, rather than as an end in itself.

Shaw is also specific about the techniques used, and their limitations:

As well as powerful neural networks, the machine learning model depends on the existence of carefully collected data from thousands of experiments. It’s still a vast screening project, just not as vast as it would be without the AI component: it uses the data to find the best ‘ready to use’ molecule from the available options.

The discovery of abaucin shows that AI is helpful for the early stage of winnowing down the vast space of chemical possibility, but there’s still a lot to do from that point onwards.

This is useful because it feels like we are in a moment where ‘AI’ is used to refer to myriad things, and using the term on its own is not very helpful. It feels akin to the early 2000s, when a whole group of technologies and applications were referred to as ‘the Internet’ (always capitalised) as though they were a single entity.

It’s notable that the abaucin study didn’t refer even once to ‘artificial intelligence,’¹ but used the somewhat more specific term ‘deep learning.’

When so many technologies, from large language models to recommendation engines to deep learning algorithms to theoretical artificial general intelligence systems are all condensed into two letters—AI—it doesn’t aid understanding. I’ve spoken to people this week who have interpreted the headlines around this to mean that something akin to ChatGPT has synthesised a new antibiotic on request—an understandable misunderstanding.

When scientists are warning about AI threatening the future of humanity, they aren’t talking about chatbots—yet you’d be hard-pressed to discern that from breathless headlines that refer to anything and everything as simply ‘AI’. In just a handful of days, even the well-respected BBC News website has published articles with headlines referencing ‘AI’ about drone aircraft, machine learning, delivery robots and image generation: all entirely different applications of a very broad class of technology.

If we’re to have sensible conversations about the ethics and regulation of AI technologies, I think there’s much to be done to try to help the public understand what exactly is being discussed. That ought to be the job of the news. Currently, it feels like we’re stuck in a cycle of labelling things as ‘AI’ as a strategy to garner attention, leading to conflated ideas and complete misunderstanding.

The image at the top of this post was generated by Midjourney, whose idea of the appearance of a human brain seems sketchier that I might have imagined.

1. Though, in fairness, the press release did. ↩

I thoroughly enjoyed the opening article in the latest n+1, in which the editors review the environment in which we currently read journalism online. Sweeping in scope and full of humour and common sense, the central conceit is that Twitter is no longer a useful platform for discovering quality journalism:

Twitter was unproductive, depressing, and a big waste of time, but until recently it never made anyone feel quite this stupid. The Musk era has been defined by a relentless barrage of idiocy, which has seeped into the infrastructure. The sense of chaos, the eternal return of memes and controversies, the algorithmic de-emphasis on tweets that link out to anything other than tweets, the emergence of the most annoying people in new and surprising contexts (Bari Weiss and Musk: a true signature collab) — all of it is too much, especially when all one wants is interesting articles.

I left Twitter years ago. I also enjoy reading quality journalism, and this article resonated with me because I, too, found myself subscribing to Air Mail and Semafor and the London Review of Books and The New Yorker and the New York Times. And I, too, found that the pub walls changed the experience, not necessarily for the worse, and I, too, wonder whether quality journalism can survive like this. And it’s interesting to read a thoughtful, detailed critical analysis of these fleeting thoughts—even while not agreeing with every one of its points.

It’s worth a few minutes of your time.

The picture at the top of this post is an AI-generated image created by OpenAI’s DALL-E 2.

There are two adverts I keep hearing at the moment. I think the combination of the two might be harmful to health.

The first says:

One in two people will get cancer in their lifetime.

The second says:

One in two smokers will die of a smoking-related disease.

I wouldn’t quarrel with either of those statements. However, I think hearing both might lead people to underestimate the risks associated with smoking. I think people think “cancer = death” and that “a smoking related disease = cancer”. They may therefore—completely inaccurately—conclude that smoking doesn’t change the risk of death all that much.

Of course, not all cancers are life-shortening—indeed, many don’t even warrant treatment, and many require only a one-off minor procedure.

Of course, there are many life-shortening smoking-related illnesses which aren’t cancers.

I worry, though, that people will just compare “one in two” with—well—“one in two.” That wouldn’t be helpful.

The picture at the top of this post is an AI-generated image created by OpenAI’s DALL-E 2.

A lot is often written about the need to adapt our national infrastructure, and particularly that of cities, to manage climate change. Most of what I read is about adapting cities to produce less carbon. For example, by promoting active travel over carbon-fuelled vehicles or by using heat pumps in place of carbon-fuelled central heating.

A lot of this is framed very poorly. Coverage regularly implies a choice, most frequently citing ’economic damage’ as a key barrier to implementing ecologically sound practice. It shouldn’t need saying that making our planet uninhabitable is the ultimate act of economic damage.

As my old friend James O’Malley frequently reminds us, adaptations actually have to be built to make a difference. If every option is blocked because it’s not quite eco-friendly enough, then we’re doomed to end humanity.

One aspect that seems under-discussed is how we need to adapt the built environment to cope with the climate change that is already baked-in through the damage we’ve already done. I was pleased to see an article in Le Monde yesterday about exactly this issue, looking at how Paris needs to adapt.

I think the two issues ought to be considered hand-in-hand: the super-insulated houses required to make heat-pumps work must also support passive cooling, or they will be uninhabitable in the medium term. Building better infrastructure for active travel is a must, but doing it with asphalt is a bad idea. Nice big windows to reduce reliance on powered lighting are unhelpful if they also trap heat.

I worry, though, that this complaint just puts me in the same category as those who oppose developments for not being quite ecologically friendly enough. I don’t think it’s the same complaint: building infrastructure in a way that guarantees a short lifespan can’t be good for the planet… but is it better than nothing?

A year ago, I wrote:

it does all lead me to wonder if the tide is turning, and whether by this time next year we’ll still see the same constant exhortations by television and radio programmes to follow their people on Twitter and to Tweet out our opinions. Maybe enthusiasm really is waning.

This was based on the toxic nature of the site and the strong evidence around its impact on mental health. And this was six months or so before Elon Musk took control of the site, a development that can’t be said to have improved either of these factors.

If anything, the TV exhortations to use Twitter have only increased. Most bulletins on the BBC News Channel now feature the presenter asking the viewer to follow them on Twitter. When I was growing up, it would have been unimaginable for a BBC newsreader to actively promote a non-BBC service from behind the desk. Now they do it almost hourly, and for free, for a service that seems to cause the BBC endless trouble.

We live in strange times.

The picture at the top of this post is an AI-generated image created by OpenAI’s DALL-E 2.

One of the reasons I, like many others, enjoy reading the FT Weekend is that it seems impossible not to learn something from every edition. Often, the lessons are unexpected.

Today’s edition has a magazine cover-feature by Henry Mance called The search for a non-toxic masculinity¹

Much of the article focuses on YouTube and its influence on young boys. YouTube is a site that I only occasionally use. I’m not up-to-date with YouTube culture, and the article’s opening question—‘Do you know who KSI is?’—won a resounding ‘no’ from me.

But quite apart from the obvious, there were three things in this article that have given me new and unexpected perspectives—and ones that I never thought this week’s FT Weekend would provide when I started reading.

YouTubers are mainly blokes hanging out with blokes. One reason is that many began in gaming, which is predominantly male. “I’ve never seen a group of girls have a YouTube show,” a 12-year-old called Rose told me. Her friend Flo added: “It’s what’s normal now. It’s so weird.”

I had no idea that YouTube was predominantly male: in fact, I’d have predicted the opposite. I haven’t previously given it much thought, but I probably conflate social media platforms, which is probably a side effect of not using any of them with any frequency.

When I think of a ‘social media video’, the heuristic which comes to mind is make-up tutorials, dance videos and self-filmed videos of women on holidays. That’s probably partly because it’s this sort of video which is referenced most often in discussions of social media. But I suppose social media refers to a diverse range of platforms, and it’s foolish of me to conflate Instagram with TikTok with YouTube. Each probably serves different needs with different audiences and different popular material.

Of course, I’m culturally aware of Andrew Tate and MrBeast and Marcus Brownlee and Tom Scott and PewDiePie as hugely successful YouTube creators.² Yet somehow that’s never filtered through to influence my view of what YouTube is, and what is popular on the platform.

Mance’s article has challenged me to think again.

In Britain, children aged between 12 and 15 say that they spend more time on social media than with their friends. One boy told me of Tate: “I feel like some of his stuff is really helpful. Some of the stuff he says is bad. It’s a separate part, that’s how I see it.” But the algorithm won’t make that distinction: it will keep serving up Tate videos, and maybe you will start agreeing with them. The influencers know that provocation boosts numbers. The algorithm pushes boys to similar, and often more extreme, content.

I’ve previously listened to an entire podcast series about the YouTube recommendation algorithm, but it was this paragraph in Mance’s article that somehow made me think differently about it.

For some utterly ill-informed reason, when I think of YouTube, I think of Google, and then think of PageRank. PageRank was the secret sauce of the original Google search engine, based on the insight that the more pages that link to a given page online, the more likely it is to be the page that someone is searching for. One of the clever things about that insight is that if you search for a subject on which opinion is split, the results are likely to be equally split: it essentially has an (obviously imperfect) inbuilt anti-bias feature—or, perhaps, a pro-populism feature.

The YouTube recommendation algorithm is engineered to do exactly the opposite: it aims to serve up videos of similar appeal to the one the user is watching. This naturally tends towards reinforcing any given set of opinions, and actively hiding the opposing view. This was the whole point of the Rabbit Hole podcast—but somehow, it’s Mance’s writing that has helped me to mentally re-categorise recommendation algorithms as something thoroughly different from search engines. I now see much more clearly why they are problematic.

I grew up before the internet was a thing. My influencers came on printed pages. Loaded magazine launched in the UK in 1994, when I was 11, with the strapline “For men who should know better”. FHM (For Him Magazine) relaunched that same year, announcing: “It’s a guy thing”.

To my generation, these magazines came as close as anything to suggesting what it meant to be a man. Loaded “captured the reality of what men were like when we were together”, its first editor James Brown tells me. Its ethos was summed up by the comedian Frank Skinner, who told the sixth issue: “I’ve never gone along with all this new man bollocks . . . I think you can talk openly about how much you like a woman’s tits without being sexist.”

Perhaps the lads’ mags ethos hadn’t faded; it had just moved online, fragmented and mutated. A secondary school teacher I spoke to drew the distinction: we grew up with casual sexism; what exists now is pseudo-intellectualised sexism. Tate doesn’t make lazy jokes about women drivers or overweight women; he preaches about the innate superiority of men and the virtues of physical fitness. And he may not be a blip.

I was nine in 1994: whether it’s because I’m slightly younger than Mance, or whether it’s just because we had different childhoods, I don’t remember these magazines having any real influence on me or my friends. Of course, we were aware of them, but I don’t think we really paid them any attention.

And perhaps that explains why I would never have thought about the continuity between magazines like that and YouTube content—nor the insidious differences between the two.

1. The online version has a different heading: ‘What does it mean to be a boy online in 2023?’ ↩
2. KSI has still passed me by, but now I know the pseudonym stands for ‘Knowledge, Strength, Integrity’ ↩

Tonight, it’s the grand final of the 2018 Eurovision Song Contest. As long-time readers will know, I really enjoy watching Eurovision – I even live blogged the UK selection programme once. There are a few reasons I really enjoy it.

Firstly, of course, the music. If I had to choose a favourite radio station, it would undoubtedly be Monocle 24. There’s quite an overlap in the Venn diagram of international music Monocle 24 would play and the sort of music that does well at Eurovision. In fact, most years, they’ve already had quite a bit of air-play of the big-hitting songs by the time the contest comes around.

Secondly, there is something so joyful about seeing so many different countries and cultures come together for a single peaceful purpose. In that regard, Eurovison is a little like the Olympics – only moreso, because the countries are peacefully scoring one another. More of this in the world would be a good thing.

Thirdly, there are bits of it which are undeniably batshit crazy. I’m not that entertained by the stuff which is out-and-out mad, but the unexpected crossovers been madness and talent which occur from time to time are quite something: take this year’s entry from Israel, which is crazy, brilliant and catchy all at the same time.

It’s this third point which makes me feel a little glum about the UK’s entries, which are typically standard, uninspiring pop fare (look at this year’s entry from SuRie). We seem to have an astounding capacity for moaning about the poor scores the UK entry receives even when the middle-of-the-road pop numbers rarely perform well even in the UK chart, despite the considerable Eurovision following. It would be really nice to have a UK entry that was quirky, whether that’s through outright craziness or just having great execution of something which is very ‘on trend’: look at this year’s entry from Sweden.

But I think the BBC lacks the boldness and creativity to find or inspire that sort of song. Whenever the BBC tries to do ‘zany’ in its programming, it tends to come off as ‘crazy by committee’ and spectacularly flops. This is even more so the case since the budget cuts at BBC Three, which was their outlet for experimental material. The best they seem able to come up with these days is crap like Don’t Scare the Hare or 101 Ways to Leave a Gameshow, which is a shame given the BBC’s lustrous history of the surreal.

The UK public vote rarely tallies with the most popular songs across Europe, even in an approximate way, so a publicly voted selection show (which the BBC has returned to using in the past couple of years) doesn’t seem like a logical way to go. Similarly, the UK jury seems permanently out of touch with the views of the rest of Europe, so professional selection doesn’t seem ideal either. I think the BBC needs to divest itself from song selection, and outsource it to people who have a chance of selecting something half decent.

The question is… who can provide that? I’d put it in the hands of the curators of the Monocle 24 playlist. They know a good song when they hear is – and have a definition of “good song” that at least approximates that of viewers across Europe.

Of course, I suspect such a system could never work in practice: I’m sure Monocle wouldn’t want to sully their upmarket brand, and the BBC wouldn’t want to lose control. But I think it’s an interesting idea!

The logo at the top is the official one for this year’s contest, taken from the press pack.

A couple of news stories I’ve read lately have made me think about our approach to regulation of advertising.

First, there was this story by Travis M Andrews in The Washington Post about the portrayal of smoking in shows made for streaming services:

Among the vices often embraced by streaming services and avoided by broadcast television is tobacco in all in its forms … A study compared seven popular Netflix shows to seven popular broadcast shows. In this sample, it found Netflix’s shows featured characters smoking almost three times as often as those produced by broadcast networks like NBC, ABC and CBS.

Now, we could spend all day poking holes in this ‘study’, but the thought is still going to fester: it does seem like there might be more smoking in these shows than in those on broadcast TV.

Second, there was this BBC Trending story by Branwen Jeffreys and Edward Main about YouTube stars being paid to encourage kids to cheat on school assignments:

YouTube stars are being paid to sell academic cheating, a BBC investigation has found. The BBC Trending investigation uncovered more than 1,400 videos with a total of more than 700 million views containing EduBirdie adverts selling cheating to students and school pupils. In some of the videos YouTubers say if you cannot be bothered to do the work, EduBirdie has a “super smart nerd” who will do it for you.

This isn’t so obviously related to health but does highlight an issue with inappropriate advertising within online streams which are typically seen by children and young people.

Both of these stories made me reflect on the work that has gone into restricting advertising of harmful products such as cigarettes and energy dense foods, and how the fruit of that work might be lost if legislation doesn’t keep up with changing media consumption habits.

For example, there are no regulations around the portrayal of smokers on streaming shows, whereas broadcast shows must comply with Ofcom’s rules, including Rule 1.10:

Smoking must generally be avoided … unless there is editorial justification.

There seems to be non-stop debate in the media press about whether TV ads or online ads are more ‘impactful’, with the conclusion usually predictable according to who has funded or published the work. But it does seem increasingly clear that many people (including me) are now watching more streamed content than broadcast content, and that this is more common among younger people.

It’s hard not to worry that the slow pace of legislative change might cause us to unintentionally slide back to an era of lesser regulation of what is actually seen despite strong evidence of harm. We really mustn’t let that happen.

The photo at the top was posted on Unsplash by Tina Rataj-Berard and is used here under the Unsplash licence.

Over my cornflakes this morning, I read Ben Hoyle’s interview with Tim Cook, CEO of Apple, in The Times Magazine.

This was one of those interviews which is sort of interesting but doesn’t really say much. Though I was quite taken with this description of Apple’s canteen where the cutlery is hidden from view in an illuminating example of form over function:

You can’t tell what the chefs are cooking because there are no menus on display (the options are on your phone if you’re an employee). You don’t seem to be able to pay cash for anything and there are no sauce sachets or eating utensils to be seen unless you know where to look (they’re with the other unsightly essentials like bottled drinks and napkins, sunk out of sight in smooth, curved central islands reminiscent of giant iPods).

What really struck me about this interview was the weird cognitive dissonance in the tenth paragraph. In this paragraph, Hoyle points out that:

Apple’s App Store is “curated” to the extent that you (and your children) won’t find hate speech or pornography on there.

That is, Apple – for better or worse – prioritises its values over the freedom of its customers to easily use the platform for activities which meet with disapproval from Apple. I wish this (puritanical?) attitude had been used to challenge in this bit of the same paragraph:

Apple has regarded privacy as “a basic human right” for a long time and “built the company around” that belief. The sprawling, intimate personal data profiles that companies like Facebook and Google compile “shouldn’t exist”, Cook thinks.

Cook claims that Apple is built around privacy. Yet, while Apple is happy preventing access to hate speech on the App Store, it actively promotes the Facebook app despite it asking for user permission to build data profiles which Cook says are antithetical to everything Apple stands for.

This seems a really odd moral position to me: if your company is reputedly built around one “basic human right”, why allow apps which violate that fundamental belief and ban apps which contravene less dearly held standards? The answer seems fairly obvious to me: the Facebook and Google apps are among the most popular, and are core to the iPhone experience. But can you really claim something is a cornerstone value if you ignore it to sell more phones?

I was also a bit riled up by this ludicrous comparison:

On cybersecurity … the company also protects its FaceTime and Messages apps with end-to-end encryption unlike, say, Google’s standard Gmail.

Why compare a closed messaging system, where end-to-end encryption is easy, with an open standard like email? That reads like a line supplied by Apple. It should have been challenged by asking if Apple’s iCloud email service protects messages with end-to-end encryption, which of course it does not.

There are a lot of things that Apple does extraordinarily well. It is evidently one of the corporate success stories of our time and has inspired phenomenol brand loyalty among a huge population of users. But it isn’t perfect.

Much of the media, and Hoyle’s article is no exception, seems far too credulous when it comes to Apple. Coverage of Apple would be much more satisfying if it showed a degree of balance or at least an attempt at challenging some of the more outlandish media lines rather than simply repeating them verbatim.

The picture of Tim Cook at the top of this post was uploaded to Flickr by Fabio Bini, and is used here under its Creative Commons licence.

I’ve mentioned Mozart in the Jungle on here before. It’s a wonderful Amazon Prime comedy-drama about passion, professionalism and music. Inspired by Blair Tindall’s autobiography of the same name, the show follows both the appointment of a new conductor to the New York Symphony Orchestra and the travails of a young oboist trying to break into the orchestral big league.

It stars Gael García Bernal and Lola Kirke who both give performances of a lifetime alongside an all-star ensemble. It is creative and imaginative to the point of being a bit nuts sometimes. What other show would have Lang Lang on as a guest star and feature his piano performance with the sound replaced by Daft Punk? And yet, this made for one of the most memorable scenes in four seasons. And the third season featured the most beautiful cinematography of Venice I have ever seen. And, of course, the whole series features fantastic orchestration spanning all kinds of music.

Mozart in the Jungle is a completely brilliant show. And yet, Amazon has decided to cancel it. I really hope someone else picks it up.