How ChatGPT could affect the workspace: the good, the bad and the ugly
According to an IBM study, 35% of all companies are actively using artificial intelligence in their day-to-day operations. The study also found that 42% are still exploring AI's potential for the future. Across the world of work people in senior management positions are putting time aside to seriously consider the ways in which AI could be used to help them work faster, smarter and to a higher standard.
That’s one reason why the sudden appearance of ChatGPT made waves. The chatbot, launched by OpenAI in November 2022, responds intelligently to human prompts and questions, from ‘Write me a recipe in the style of Jane Austen’ to ‘What is the meaning of life?’ Unsurprisingly, people couldn’t get enough of it. LinkedIn and Twitter feeds were filled up with ChatGPT’s responses to a wide variety of human inputs. And the consensus seemed to be that this was the ‘tipping point’: AI had officially landed.
We can think of how ChatGPT could affect the workplace. It could act like a virtual assistant, helping us to organise our schedules. It could respond to certain emails or create basic email marketing campaigns. It could be used to create first drafts of webpages, to be neatened up later by a copywriter, or provide prompts for team brainstorms to help get a conversation started.
Steven Piantadosi has shown, with a little bit of creativity ChatGPT can be made to say highly problematic things. For example, when asked for a list of the most valuable brains, broken down by race, ChatGPT returned a table with white brains (worth $5,000) at the top.
There are safeguards built into ChatGPT, but when these are bypassed – and they can be bypassed – we find that ChatGPT is not immune from the bias that affected voice assistants, for instance, or web searches, or facial recognition technology. Voice assistants, which have often reflected stereotypes of submissive female secretaries, have more difficulty understanding female voices than male ones. Even gender-neutral web searches often returned male-orientated results. And facial recognition’s accuracy is lower for women and people of colour than it is for white males
Evidently, AI is not quite ready to play a major role in the workplace, and certainly not in diversity, equity and inclusion (DEI). The technology itself is outstanding but if it is biased, then bringing it into the workplace risks entrenching tired and harmful stereotypes in subtle ways that may not even be easy to detect. While ChatGPT could, in the long term, prove to be a vital tool of hiring experts, helping potential recruits to overcome disadvantages to produce strong CVs or surmount cultural or linguistic barriers, it could, in the short term, perpetuate the kinds of problems that we’re trying – steadily but successfully – to overcome.
The solution is, of course, more diversity. The more diverse the teams that build and improve AI, the more that AI will reflect that diversity. What we want is AI that is, above all, human in its outlook. Not a specifically white, Black, male or female human, but a human. And for that to happen, we’re going to need to start paying even greater attention to representation in tech. WISE has found that just 21% of IT professionals and 12.5% of engineers are women, compared with more than half the population. And a survey at our Women of Silicon Roundabout event last year suggested that seeing and hearing from other women in tech was a major attraction, which underscores their absence at the highest levels in the sector.
The good news is that awareness of bias in ChatGPT and other forms of AI could provide us with the motivation we all need really to accelerate change in tech and increase diversity across the board. We all want to make use of the amazing developments in the space to make our lives easier and address the challenges we’re facing. If the condition for doing that effectively is to increase diversity, then we might just see companies investing heavily in doing that. There is a huge opportunity for the team that can eradicate bias from AI.
For the moment, however, we might want to hesitate before we involve technology like ChatGPT too heavily in our working lives. The progress towards DEI that we’ve made over the past few years has been fantastic. Biased AI could just undermine it.
Michaela Jeffery-Morrison, pictured below, is CEO and co-founder of Ascend Global Media