Antonym: The Fake Work Edition

Let me ask ChatGPT that for you.

Dear Reader

The sun is out, Glastonbury and Wimbledon are about to begin, democratic regime change is afoot in the UK, and most wonderfully of all… the FT Summer Books List is out. I have some book recommendations of my own below.

But first: disappointing writing...

Lazy shallow-fakes 

Deep-fakes are scary, but shallow-fakes are just plain irritating.

In a recent BN Edition newsletter, we wrote about “slop” :

Slop is an old word that has been repurposed for a modern curse: carelessly created AI content.

Appropriated to describe low-quality, AI-generated content made to generate ad revenue and grub a little SEO advantage, “slop” can also be applied to generic, poorly made images (the new clip-art), and to overly-long emails that obviously weren’t written by a human.

It's not just spammers and SEO claques abusing the ability of AI to generate plausible simulations of human writing. Speaking with people in firms where they are encouraging people to use generative AI, the number one bugbear is colleagues sending obviously AI generated text in emails or reports. 

"By all means use AI to structure a report or longer email," said one executive at a recent workshop, "But if it's obviously all been written by a bot my heart sinks. I know they haven’t bothered to check it or think about it."

Using AI thoughtfully says you're smart. Careless cut-and-pastes make you look lazy or dumb.

Back in the Noughties there was a brief office craze for teasing colleagues who asked obviously simple questions with the use of Let Me Google That For You. The questioner would get a link from a colleague which would open to reveal an animation of their question being typed into Google and the actual results appearing. 

The 2024 equivalent is asking someone for their advice and getting a a response that sounds like a US-based content factory wrote it, completed with "z" where "s" should be and the following red flags (for me, at least):

• “Advancements” - we say "advances" as a noun, but AI loves to say advancements.

• “Fostering” – when they mean creating, developing or encouraging. 

• “In the realm” – typical padding that AI

• "In conclusion" — who has had cause to write that since they finished taking essay-based exams?

Spotting AI text: You can’t, but you can. 

AI cannot spot work by AI. It will tell you it can, but it can’t. “AI-detection” tools cannot reliably do it either. They give a probability that we read as a certainty.

“60% chance it was AI generated? I knew it! Fire them!”

The irony of AI spotting-tools is that they use AI. Human intuition – instinct based on years of using language – will also spot the same patterns and clues as the AI detectors. AI and humans are both often wrong. 

False positives are far too common. That bland, wordy mess might just be good old fashioned bad or bland writing. (NB: they also may penalise non-native English speakers.)

If an employer or educator suspects that someone has used AI instead of doing work they wanted their human brain to do, the better course of action is to talk about why the writing is poor, and even more about why prose that feels like it was written by an AI will lose them respect or readers. 

Speaking of accusing people of doing the wrong thing...

How to save thousands on employee surveillance

Is this really how managers think WFH works?

Apparently, yes. Wells Fargo in the US just fired "more than a dozen" staff who said they were working from home, but who "simulated keyboard activity".

We talked about this WFH vs. Big Brother battle two years ago in Antonym: The Mouse Jiggler Edition - mouse jigglers being those devices that keep your mouse moving and so your screen stays active while you aren't at your home office desk: 

You just know that this is an arms race between IT and individuals, don’t you? Somewhere, there are teams of software professionals working on apps that will spot a mouse jiggler. One day, there will be an employment tribunal or court case where a red-faced employer will have to explain that their surveillance software detected a mouse jiggler faking an employee’s presence in front of a laptop. 

This is all a massive distraction from real workplace problems. 

Firms pay a very high price for not trusting workers. Firms that can't trust their staff have to pay for surveillance, lawyers and untold hours of executive time in an effort says: "We believe that the only real work happens when hands are on keyboard and eyes are on screens. We have no other way of judging the value of our employee's efforts than their presence."

Public Service Announcement: If you have well-trained competent managers, you can measure outputs and progress to actual goals. Save thousands on those mouse-jiggler detectors.

And yes this is an affiliate link to buying a mouse jiggler if you want one. Though if you need one I would also suggest looking into a change of job. For the record, Antonym is an employer.

Deepfake fun: Kier and Nigel’s streaming war

Source https://vm.tiktok.com/ZGegWrARL/

This week’s recommendations 

Reading…

The Rehearsal, by Eleanor Catton. I enjoyed the writing in Catton’s Birnam Wood so much I had to just start reading something else by her immediately.

The Playbook, by James S Shapiro. This is a history book about today’s US culture wars, with obvious relevance for the rest of our populist-haunted world. It looks at the first successful Congressman Martin Dies’s innovations in his successful campaign to shut down the Federal Theatre, a New Deal project that brought theatre performances and social issues to millions of Americans in the years before the Second World War. His way of waging culture war set a patten that is still used today.

Making Sense of Chaos, by Doyne Farmer. Farmer is a pioneer of the increasingly relevant field of complexity economics. It seems incredible that complexity economics is not mainstream already, but it may yet become the norm. Eric Beinhocker’s The Origin of Wealth is more than a decade old, and we’ve been through several crises since that show the weakness of old models. This book is essential reading whether the markets or the climate crisis are top of your priority list.

Watching… 

Tokyo Vice. (BBC iPlayer) The second series of the Michael Mann-produced series based on a 90s journalist’s experience investigating the Yakuza organised crime organisations is top notch. It doesn’t have the urgency of the first series, but the characters and 

The Big Cigar. Apple TV is reliably high production-values, and this high octane telling of the true story of Black Panther founder Huey Newton escaping a manhunt with the flamboyant and coke-fuelled Bruce Schneider (legendary producer of Easy Rider) is a classy product. 

The Boys. (Amazon Prime) Definitely not to everyone’s taste, given the intense and frequent sex and violence, but this is one of the best long-running TV series to emerge from the streaming wars era, with its conceit of a populist-era America playing out its culture wars in a world where superheroes are numerous, commercialised and awful. 

That’s all for this week…

Thank you for reading. Throw us a like or a share if you made it this far and found something you liked.

Antony