When seeing is no longer believing: AI glasses enter the room
After a Reform UK politician’s suspected use of smart glasses as a teleprompter in the Senedd reignited the debate over wearable technology, Paul Harvey asks whether the events industry is prepared for the trust implications.
In August, I wrote about ‘pervert glasses’: AI-powered smart glasses that look like ordinary eyewear but can potentially record what the wearer sees and hears.
Now, a political row in Wales has brought the technology back into the spotlight - and raised another question for the events industry.
Jason O’Connell, a Reform UK member of the Senedd, has faced scrutiny after fellow politicians noticed flashes of light on his glasses while he delivered a speech in the chamber without referring to notes. He was suspected of using a teleprompter feature that displayed text through the lenses.
There has been no suggestion that O’Connell was recording proceedings, and Reform UK said the glasses did not have a camera. The incident has nevertheless prompted calls for clearer rules around smart glasses, with similar concerns raised at Westminster.
On one level, this might seem like a story about a politician finding a new way to read a speech. But it also demonstrates just how quickly wearable technology is moving into situations where its capabilities - and its presence - may not be obvious to the people around it.
The problem isn't just recording
Technology that can deliver prompts directly to the wearer raises questions about how we interpret what we see and hear. Is someone speaking from their own knowledge? Are they receiving real-time assistance? Does it matter, and should anyone else need to know?
If a politician can deliver a speech while reading prompts displayed discreetly in their lenses, the same technology could allow a keynote speaker to access notes, recall statistics or receive real-time assistance without anyone in the audience knowing. A presenter could appear to be speaking entirely from memory while following a script. Someone answering a difficult question could, in theory, be receiving AI-generated suggestions as they speak.
Is that necessarily a problem? Not always. Speakers have long used teleprompters, monitors, and notes – and I’m not for a minute suggesting they shouldn’t. Nobody is under any illusion that a speaker delivering a polished keynote is necessarily doing so entirely from memory. (Although, it bears saying that it’s always impressive to see someone speaking genuinely without notes.)
However, there is a difference between using technology openly as a tool and using it in a way that changes the audience’s understanding of what is happening. A teleprompter on a lectern is there for everyone to see. Discreetly reading a script from a pair of glasses, while giving the impression that you’re speaking entirely from memory, feels different. It feels dishonest, as though you’re trying to pull the wool over your audience’s eyes.
There is nothing wrong with needing notes or a prompt, but there is something questionable about going to lengths to conceal the fact that you’re using them, particularly when the impression you’re trying to create is one of effortless expertise.
Ultimately, trust is difficult to build and remarkably easy to lose. If we want audiences to trust what they hear, it’s important we don't use technology to pretend to be something we're not.
The main lesson I would draw from the Senedd incident, then, is that honesty and trust are paramount, and no one likes to feel they’re being lied to. But surely we knew this already?
AI technology itself may be new to us, but so many of the ethical dilemmas it creates are not. Is honesty a good thing? Should speakers be transparent with their audience? These are not difficult questions.
The problem is that AI is seductive; the technology makes it so easy to think we can get away with pulling a fast one. The real challenge of our new favourite plaything, perhaps, is knowing when it's appropriate to use it - and not giving in to temptation to use it when we know it's not.