When are the robots coming for us? Using AI in the workplace

Since the hype around Chat GPT first emerged there has been lots of discussion around the role of artificial intelligence in the workplace and if/when we will get replaced by our robot overlords.

robot A robot army

We've recently been auditing our videos to see which transcriptions have slipped the net and reviewing our processes to be and do better in future. It's given us the opportunity to look at the gap between automated transcriptions and the efforts of transcription companies. The standards of automated transcription have got a lot better over the last couple of years, but they still need a human eye to check them.

Tech colonialism

One of the issues with automated transcription is how it perceives non-English names. AI is currently providing an anglocentric view of the world because of the data it's working with. It generally works for our videos because they're produced in the English language. But whilst having my name changed to Derek is quite funny initially, the novelty definitely wears off. I'm regularly reminded of this flaw whenever I use mapping systems back home – they butcher Welsh placenames as a matter of course. Projects like Mapio Cymru and Common Voice are so important in addressing this.

There are also issues with how AI Algorithms are seeing and representing the world as stereotypes. AI might be able to augment human outputs, but if we're not triangulating those outputs with other sources, we're letting the biases beneath them go unchecked.

Taking a different perspective

Lots of the proposed cases I've seen for AI have been as a labour saver, but that labour is just displaced somewhere else. This article on how people in Nairobi are processing data to train AI is depressing – our exploitative use of AI in the West is in turn leading to the exploitation of people in the Global South.

If we take an extractive perspective where we contribute nothing and take value, we are in trouble. But if we look at how we can work in partnership with technology, then that opens up a whole world of possibilities. This article by Janet Vertesi delves into what a healthier approach would look like and why it's more likely to succeed. For example Spotify and other streaming services are cannibalising the music industry, but if we looked at algorithms as a source of creativity instead, then the system feels a lot healthier:

“Artists could write or curate their own algorithms to fuel creativity and retain credit for their work. Of course, rejecting replacement does not eliminate all ethical concerns with AI. But many problems associated with human livelihood, agency and bias shift when replacement is no longer the goal.”

This reminded me of Vanguard's work on public service, where they talk about the importance of starting with changing our mindsets before tinkering with systems and performance. If our thinking dictates that AI is solely a means of saving money, then our defacto purpose becomes delivering a cheap service, not a good one. Technology isn't an answer to the questions that we face, it is a tool that can help us to do better things. And if we approach AI with that perspective, then we will be much better placed to deliver human-centred services and support.

Follow me on toot.wales