“Trish, do u think AI will take all of our jobs? I’m kinda scared…”
Hi there, and welcome back to another week of Ask Trish! I hope you’re all well and having a wonderful start to April.
Thank you so much to this week’s question-er for the fantastic question. It’s a question I think we’ve all – adults and youth alike – been wondering lately: Will AI disrupt our livelihoods? The short answer is yes, AI will almost certainly take over certain tasks we currently perform and in so doing, may affect/alter some jobs. But there is a longer answer, too, and that longer answer can best be summed up as: it depends. Namely, the impact of AI depends on i) what types of AI you’re thinking of, ii) what types of jobs you’re thinking of, and iii) the timeline you’re thinking of. In this week’s post, I’ll offer you all an overview of those three factors and how different assumptions may change our answer to this question. Hopefully, this will leave you with a more nuanced understanding of AI’s potential influence.
Sound good? Let’s dive in:
First of all, our answer to this question depends on the type of AI you’re thinking of. Computer vision AI, for instance, which is able to recognize and categorize different objects, will likely have far less of an impact compared to flexible large language models (LLMs) like ChatGPT. As researchers at MIT have argued, not only are the tasks that computer vision AI can address relatively limited, the fact that these tasks are so small and fragmented makes automating them unreasonable. After all, is it really worth automating a check for whether a hospital tray – or a restaurant table – has everything it needs? A human can do that very quickly, and at low cost. But LLMs like ChatGPT go beyond simply identifying objects or completing easily defined tasks; they can actually perform unstructured, open-ended tasks – the types of tasks we do, tasks that usually require immense amounts of human cognitive labor. That they can replace this cognitive labor, with the power of tons and tons of data – and complete these tasks far more quickly – makes this type of AI potentially much more disruptive. It’s no wonder that OpenAI recently concluded that “that approximately 80% of the U.S. workforce could have at least 10% of their work tasks affected by the introduction of GPT, while around 19% of workers may see at least 50% of their tasks impacted.” Of course, it’s important to note that all that LLMs’ disruptive potential is limited by its own limits, e.g., AI chatbots have been well-known to spit out false, unreliable information, sometimes referred to as hallucinations. So even if LLMs are better poised to disrupt jobs, so long as they continue to be unreliable, it’s not clear that they will…
Similarly, the types of jobs you’re thinking of matter too. Jobs that require cognitive labor that are still out of reach for LLMs are (at least, currently) safer, whereas jobs that can be easily replaced by LLMs are far less safe. Those include jobs like copywriting (ChatGPT can write pretty well) and serving as an assistant (many AI companies have pitched LLMs as being future mini-assistants that we can all rely on). It’s also important to remember that there may be certain jobs we will never want AI to do. Already, the use of technology and algorithms in the justice system, for instance, has drawn scrutiny because of the biased, prejudicial outcomes they can produce. It’s also difficult to hold an AI technology accountable, which may make us wary of employing these technologies in situations where accountability matters. For instance, if a doctor operating on us messes up, we can sue them. But if it’s an AI technology performing the operation – and they mess up – who do we hold accountable? The hospital? The doctor observing the operation? What if there was nothing either of those parties could have done? In light of these thorny questions – and the seriousness of these contexts – some have called for limiting AI’s use in these spheres. That is, AI can perhaps be deployed in the health and justice sectors, but only in specific, limited ways.
Finally, the timeline matters too. AI is very unlikely to drastically impact our jobs in the short-term, but they will likely transform jobs in the future. How that’ll happen, though, remains an open question – in fact, thanks to the fact that AI is likely to make gradual, rather than immediate, changes, policymakers and governments around the world also have time to think through in what ways they’d like (and not like!) AI to impact society. That includes how we work! So while AI will, over time, make its mark, we humans are not powerless, either. In fact, we’ll likely be shaping its direction and influence over time.
Hopefully, that gave you a good sense of the ways in which AI may or may not impact our livelihoods. No doubt, AI will be transformative…but don’t forget: we have the agency to draw lines where we want to (and hopefully, where appropriate, we will!). Do let me know if you have any other AI-related questions on your mind! In fact, if you do, please go ahead and share those thoughts (or whatever tech or internet-related concerns you have) here. I’m excited to learn more about what’s on your mind, and hopefully, keep sharing helpful info. Thank you in advance for contributing!
Have a great week,
Trish