Who gets to say you're telling the truth?
|
Platforms are moving to retain trust. Broadcasters are declaring themselves "trustworthy" Read time: 5 mins Hi Reader, It once helped you to "write with AI" and now it's asking you to help identify AI slop. It's a sharp move by LinkedIn and a good read of where digital culture is at with low-level AI content. It might look like a way to endear themselves to the increasingly influential millennial and Gen Z age group, but there's a business reason in it too (of course). I think they're starting to recognise that the most valuable currency is trust. They have to protect it to sell it to advertisers. The platform everyone loves to mock seems to be maturing (or maybe reinventing is the word). It used to be somewhere you go to look for a new job, to announce your layoff, and then you forget about it whilst you don't need it. (Hands up if that's you too.) The content has shifted since then. People post selfies, personal stories, professional opinions, and sometimes even jokes. I've run various content sprints on my own LinkedIn over the last year, and the posts that "perform" well are the ones where I didn't perform at all. It worked best when I used it like an old Facebook account or messaging in the group chat. Every other platform has fatigued its user base, especially with the apparent obligation to post videos. LinkedIn has struggled in that battle too. But people still show up there to read, learn, and do business. Only about 1 in 100 people actually post, so anyone who does lands in front of a large audience with little competition. It isn't as daunting as it seems. You don't need a huge following or content that looks expensive. Just a point of view, and enough consistency that people know what to expect from you. Here's where I think the button runs into trouble. Last winter I surveyed 322 people across the Bages about their digital habits. General public, all ages, not a marketing crowd. Over half rated themselves 4 or 5 out of 5 at spotting AI content. But ask what they can actually identify, and it's video and images. Text comes in at 26%. The one format nobody can reliably detect is the one LinkedIn has just asked us all to report on. We trained our eyes on six-fingered hands and that plastic sheen. Nobody trained their ear for a sentence. What the button will catch is writing that feels slightly off, which describes most people writing in a second language, anyone taught to write formally, and anyone whose post annoyed you that morning. A report button is easy to aim at an opinion rather than a bot. There's a commercial reason for all this. LinkedIn is also looking to pitch advertisers on its authority in the fast-growing AI answer engine space as brands step up efforts to boost visibility on platforms like ChatGPT, Claude and Gemini. Advertisers are also recognising that the majority of buyers of business products and services are now in the millennial and Gen Z age groups, and they are looking to trusted sources like creators, 1st and 2nd degree connections when making their decisions. My survey says the same from the other end. Asked what content they enjoy, people chose creators they follow and people they personally know. Brands and companies got 10%. So to really win the war against AI slop, LinkedIn should reward people for posting good content. Nobody is incentivised to make the effort right now, so they tick the box instead, and the shortcut is having AI do it for them. Substack introduced a tool recently that measures how much of a piece is AI-written. There's something in that idea. LinkedIn could limit reach on posts with a high percentage of AI-generated content, and increase it on posts with a low percentage. Imperfect. Those detectors misfire on second-language writers, which is most of the people I work with. AI slop is only going to increase in volume. The best way to beat it might be to go hard on rewarding the people willing not to take the shortcut. Two more numbers from that survey: 64% feel worried or deceived when they discover something was made with AI, and 71% of those people use AI themselves. Where I'm at, for now, is that I'll use it to argue with a draft and find the blind spot. I won't use it to have the thought. What are you currently outsourcing that you'd be embarrassed to admit was outsourced? Escape the algorithm Tool: Pangram. The AI detection tool being backed by more and more tech companies. NEWS: I nearly posted that the BBC and ITV had gone begging the government for priority in your feed. I was wrong. It's the government's idea. DCMS put out a Green Paper in June asking whether social platforms should be made to push public service news up your feed and search results. Voluntary first, legislation if that fails. Consultation closes 31 August. My take: they didn't lose our attention because people couldn't find them. Three-quarters of 16-24s already get their news on social. They found it and scrolled past. You can legislate a slot. You can't legislate someone caring. Reel: Grace Andrews - The government wants to control what you see on your social feeds RELATED: the Cabinet Office is hiring a stack of social media people (12 of its 31 open roles, £57-68k each.) Doing the rounds as "Burnham's building a TikTok team." Not quite. The Government Media Unit was set up in April, before he was PM, and it's civil service, so it's government comms, not his personal brand. Still, what we can assume from this is that the state has decided press releases don't reach anyone anymore and is building its own content studio. Sits funny next to the Green Paper asking platforms to push trustworthy news up your feed. One hand asking for the slot. The other hand learning to earn it.
So, to summarise: LinkedIn added a "seems like AI slop" button, the UK government asked platforms to push "trustworthy" news up your feed, and the Cabinet Office started hiring 12 content people to publish straight to citizens. All three are chasing the same problem. Nobody knows what to believe anymore, and all three reached for distribution instead of trust. Watch who ends up deciding what is "trustworthy" and "authentic", because whoever does can hand it out and take it back. If you got this far, you're part of 71% of subscribers, so thank you, Reader.
If this was forwarded to you, you can subscribe here. See you in two weeks. Peace, Has
If something in this issue sparked a thought about your own content, I'm easy to reach. Tell me what you're trying to achieve. |
