Generative AI and the mediocre worker
Jon Boys discusses the benefits of generative AI tools, and how organisations can utilise them
Jon Boys discusses the benefits of generative AI tools, and how organisations can utilise them
‘Mediocre’ can be used pejoratively but it shouldn’t be. It simply means ‘in the middle’ (medi) which is where most people (by definition) reside. Indeed, a great society would be one in which the mediocre (and below) performer can thrive, and I believe AI has a big role to play here.
Take for example writing tasks, one of the few applications most people have discovered from ChatGPT. Writing ability and vocabulary size increase with age. While age 30 might be considered old for a tennis player, 40 is considered young for a novelist. Writing can be improved of course but the sheer passage of time will introduce one to millions more words and sentence structures which will become internalised to one’s own voice and style. Younger people have had less exposure to language, and therefore will tend towards the mediocre end of the spectrum. AI can smooth over this by rewriting content more cogently and in any style you ask it to. We are seeing this already – for example, with HR practitioners reporting a high volume of well-written job applications. It's now more difficult to do the first sift because all applications are well-written.
Writing is the tip of the iceberg. To my surprise, and frankly horror, ChatGPT’s code interpreter can help a layperson write better code on day one than I could write after years of practice. When asked to clean up a spreadsheet, chart the data, and provide insights, the outputs produced were on par with those of an analyst with some years of experience.
Traditional wisdom holds that automating these tedious tasks will free us up for more valuable work. What is less talked about is how this AI puts more valuable tasks within reach of less skilled people and these people may be the biggest beneficiaries.
Tacit knowledge is supposed to be the quintessence. That extra thing that can’t be codified. It’s the knowledge sharing that happens through socialisation and experience, or ‘water cooler’ moments.
AI can take something, for example a chunk of code, and put the cues around it to make it accessible to a specific task of yours. It’s almost like it’s taking the output and reverse engineering the tacit knowledge needed to produce that output. My experience of AI leads me to believe that much of what we thought was tacit knowledge can in fact be codified. For me this is evident in those deus ex machina moments, or what technologists call ‘uncanny valley’, when a machine gives the illusion of consciousness (these also tend to be the lightbulb moments that convert people to the use of AI in their daily work). At these moments the pool of what we considered to be tacit and human gets smaller.
The beauty of AI is not so much facilitating knowledge within the organisation as leveraging knowledge from outside. That knowledge always existed somewhere out there in the ether of the internet which was ultimately used as training data for the model. But without the conduit of a Large Language Model (LLM) to organise it for us, it was not so helpful. AI chatbots can be quite forgiving, in a way people and traditional search engines are not (here, I am referring to the traditional role of a search engine putting you in touch with human generated content. Search engines increasingly use AI and have chatbot features such that they may be indistinguishable in the future). If you don’t use the right term, the AI can usually work out what you mean but search engines would not. People often bid for the SEO of tangentially related content making sifting through Google for a solution tedious.
AI leverages knowledge from outside the firm. There is firm-specific knowledge like the colour hex codes used in a company's brand guidelines, or its tone of voice (although I’m sure we could get AI to read a company's guidelines and internalise the rules) but my experience of using AI suggests that most problems we have in our own organisations are just modules of other problems. The implication of this is that firm-specific capital, a sort of individual asset (ie something that belongs to someone from which they can derive a flow of benefits), will reduce in the age of AI. A terrifying prospect for those with long tenure but a boon to new starters.
In pre-ChatGPT times, it was popular to say that to maintain an edge we should focus on the things that computers will find difficult to do, like creativity and empathy. These sound like platitudes in the context of the new machine age. Generative AI brings empathy and creativity to the masses. Anyone who has interacted with ChatGPT will know it has exceptional soft skills. It’s very patient with me when I continue to get something wrong and always allows me to save face.
And the creativity is something to behold be it with text, images, or code. I have been using LangAI as a substitute for a language exchange partner (it’s not fussy about time zones). I recently asked what its favourite tree was and it answered, ‘the weeping willow, because its branches fall graciously, and it has a melancholy look’ (it sounded even more poetic in Spanish). I was a little saddened that the conversation was perhaps the best one I had had that day. I’ve been asking people what their favourite tree is and I’m yet to hear a better answer.
What we are seeing then is a democratisation of different types of skills, some of which used to take many years to acquire, including writing, coding, drawing (I made the image below using generative AI programme Midjourney), creativity and empathy. And perhaps a lessening of the importance of firm-specific capital. All this could pave the way for the mid-performing worker/new starter. Given that this includes most workers, this could be good news. From a firm’s point of view, if the same outputs can be produced with a lower level of human capital, then this could potentially solve skills shortages in organisations. This is particularly attractive in the context of current tight labour markets.
Art that I created in Midjourney using the prompt “an office worker who is happy at their desk using generative AI to be productive”.
We may even see a reversal of the superstar effect, which describes a winner-takes-all scenario for the best firms and workers who are able to exploit technology. This results in increased wage inequality. The upshot would be a fall in wage inequality as the superstars become more substitutable.
As ever with AI there is the spectre of mass unemployment. Something I’ll save for a future discussion.
Browse our A–Z catalogue of information, guidance and resources covering all aspects of people practice.
Discover our practice guidance and recommendations to tackle bullying and harassment in the workplace.
Jon joined the CIPD in January 2019 as an Economist. He is an experienced labour market analyst with expertise in pay and conditions, education and skills, and productivity.
Jon primarily uses quantitative techniques to uncover insights in labour market data, both publicly available and generated through in house surveying. Jon regularly contributes commentary and analysis of economic issues on the world of work to online, print and TV media. Recent work includes the creation of an international ranking of work quality, analysis of firm level gender pay gap reporting data, and an ongoing programme of work looking at the changing age profile of the UK workforce.
We examine and outline recent research investigating the impact of generative AI tools on the HR profession
Read our submission to the Irish Government on the implementation of the EU Artificial Intelligence Act 2024
We look at how investing in digital technologies, HR skills and culture drive success in restructured people functions
Browse our resources on the use and impact of artificial intelligence technology in the workplace
How can people teams balance line managers’ need for operational people management support while growing their team’s strategic influence through the HRBP role?
We examine and outline recent research investigating the impact of generative AI tools on the HR profession
We look at how investing in digital technologies, HR skills and culture drive success in restructured people functions
We look at the main focus areas and share practical examples from organisations who are optimising their HR operating model