With reports about how around 18% of all human work could be automated by AI, there’s understandably some panic around whether AI will take your job, hereby rendering you completely useless as your only function is to contribute towards GDP growth, in the same way a bee is essentially just a vessel to create honey.
I am here to partially assuage your fears and partially to freak you out even more. Or maybe I’m not even here, and I’m actually an AI designed to contribute bottom of the barrel blogs based on trending content. Anyways, let’s dive in :
The world of work is where the biggest disruption is slated to be coming, with roughly two thirds of jobs across the UK and the US to be exposed to some form of automation, and even more pertinently, 7% of US jobs will have over half of their tasks automated, leaving them at severe risk of replacement.
AI is really gunning to replace desk jobs as it hasn’t been plugged into a corporal form yet, presumably due to a bunch of PostIt notes at the OpenAI offices imploring workers “DO NOT PLUG AN AI BRAIN INTO A ROBOT” with a copy of Terminator 2 constantly playing on loop.
To work out if I am at risk for automation, you need to know what I do and that’s something a large portion of you probably don’t know. Broadly my job consists of three main things :
Writing, building and analysing paid advertising
Writing content and optimising websites so they appear on search engines like Google in high positions
Managing websites with A/B tests, data analytics and a bunch of other tools to make sure the website is functioning in the way it should be.
The bad news for me is this is stuff AI is already really really good at. It can write content incredibly well and quickly (if a bit blandly) and also can undoubtedly work out how to set up a Facebook or Google ad, I learned when I was 22 and I’m pretty sure I would have failed the Turing test at that age.
That being said, you may be quite surprised to find that I’m not super worried about the impact of AI on my role, and my reasons are twofold.
The first is that I view AI as a tool, rather than a replacement. Generative AI only produces excellent results when the prompts are specific enough to warrant an excellent result. Given the above knowledge of my area of work, could you write a prompt that produces the same results as me? Maybe, but could you write a prompt that produces the same results as a prompt I can write? Absolutely not, unless you are my boss, in which case, hi Curran! The knowledge I have still plays to my advantage, in the same way an accountant or data scientist can probably make a way better (and more valuable) spreadsheet than I can, despite the tools being freely available to both of us. And that leads me to the second reason that I am not too worried about AI.
Did you know, in 1890, 1 in 12 adult US men were employed in the rail industry. I don’t know if you have checked recently, but that’s not the case anymore, because cars came along and suddenly the only mode of transport aside from sitting on an animal being a metal tube filled with other people going kind of near where you need to go didn’t seem like a great idea anymore.
Massive technological changes don’t just strip jobs, they usually create them as well. There are probably going to be whole swathes and industries created from this advancement that we can’t even predict (who in 1890 would have predicted that there would be companies that make whacky arm inflatable tube guys).
A really good insight into this was actually from a podcast I was listening to by the founder of OpenAI, Sam Altman, who shared that despite an AI being able to easily beat a grandmaster in chess since the late 1990’s, chess has never been more popular. Just because an AI can do something, doesn’t mean that it will automatically take the place of a person doing it, and it’s more likely that it will do it in conjunction with the person.
It’s a tool and you can’t afford to not know how to use it.
The key is to remain informed and embrace AI technologies in your role. I can’t recommend the book How To Talk To Robots enough for this, as it does a great job of explaining how non-technical people can utilise and understand artificial intelligence tech. If you are interested more in what I’m reading, you can check out this blog I wrote about my top books last year.
Small caveat here is the freelance economy : I do worry about freelancers (especially creatives) ability to entice new customers as there will (and undoubtedly already is) an attitude of “fuck that we will get ChatGPT to do it”. My advice to freelancers would be to lean into using AI and try to focus on real world results of your work. I guarantee that a skilled graphic designer can create a more valuable end result than Midjourney for example, but it’s about proving that difference. Stay informed, stay learning and good luck!
I like to code. AI can code. Pretty well. And that’s concerning to me because I’m like 30 hours deep into some projects at the moment and it would be incredibly depressing to feel like it was all for nothing.
However, ChatGPT has also never made coding easier and more productive. I can ask simple questions, ask for advice and get inspiration for new features and improvements. It can even help with basic debugging which, with my code looking like someone threw a bunch of fridge magnets at a wall, is incredibly impressive.
Now, it’s important to note that all of this information was readily available to me (and everyone) before ChatGPT came out. Google is so prevalent and everyday that I think the magnitude of how amazing an index of pretty much all human information is has become underappreciated. Generative AI is just all of this information squished down and presented in a chatbot, and is much much quicker which is amazing. It also has the ability to suggest and mimic human conversation which can be really helpful when you have spent hours coding and forget how to talk normally.
I think my passion for coding is such that even if the AI can code better than me, I’m still going to do it, in much the same way people still make handmade chairs despite a robot being able to churn out thousands in the same amount of time. So I’m not super worried about AI impacting my personal passion projects, although it probably will impact my ability to make money from them.
Now you may have been tricked into thinking I’m sipping margaritas on the beach, relaxed as anything about the impending AI boom, but that’s where you are wrong. Remember the bit in the intro when I said I was here to relax you and also freak you out, well here comes the freaking out bit.
I’ve written before about tech’s problematic obsession with a fantasy meritocracy, and the impact that this wilful ignorance has on minorities. Generative AI is trained on vast swathes of data that contains the very same biases and prejudices that I wrote about in that (excellently written if I may so myself) blog above. As such, it replicates these biases, and this can be disastrous given the rapid speed at which we are plugging AI into literally everything. What I’m basically saying is that if we aren’t careful we are going to end up with racist toasters and misogynistic parking meters.
Don’t believe me? In Coders, Clive Thompson has a whole chapter dedicated to the ways AI has previously exhibited extreme bias in its responses due to its training, including but not limited to :
Labelling black people as gorillas
Giving Mexican restaurants a lower rating than any other type of cuisine
Only showing adverts for high paying jobs to men over equally qualified women
Giving black defendants a higher risk of recidivism score over white counterparts when all other factors were controlled for
Now, these were early stage AIs that were being examined, and GPT4 and the slew of other AIs that have come out are much more advanced, but there has still been nowhere near enough research and safeguarding put into place to stop this type of behaviour from existing. In fact, it's going to be almost impossible to control, as the data needed to train larger and larger models will exponentially increase, meaning proofing and vetting the content will be next to impossible, especially when AI is being used for purely private gain by companies who rush it out the door to make profit.
Imagine a biased AI being put in charge of loan approvals or credit scores or writing complex legal contracts.
It’s often touted that AI will help usher in a new era of humanity, but unless we are super careful, we are going to bring things from the current era that should have been left in the previous era anyway.