In 1779 Ned Ludd, a weaver living in Leicester, broke two stocking frames in a fit of rage. The frames, designed to replace traditional hand knitters, had become emblematic of the technological development that was sweeping Britain, making centuries-old practices redundant. Years later, Ludd’s story would be repurposed into the folkloric character of Captain Ludd, inspiring the Luddite rebellions and resulting in the destruction of textile machines, which were seen by workers as threatening their craft.
It is debatable whether Ludd actually existed, but his status is a recognition of the sense of helplessness workers in industrial England felt in the face of rapid technological change. They saw their only source of help as coming from a mythical figure.
Reflecting on recent developments in Artificial Intelligence, I feel a similar sense of helplessness. The success of ChatGPT, an artificial intelligence chatbot produced by OpenAI, has showcased AI’s power and scope. The Chatbot was estimated to have reached 100 million monthly active users in January, just two months after its launch, making it the fastest growing consumer application in history. Trained from data found on the internet, it seeks to understand and respond to written prompts. In this regard, it’s incredibly good: it can craft resumes, solve maths problems, and even help to write code. Its successor GPT-4, able to respond to image prompts, is even better. One user showed a photo of the inside of a fridge to the bot, enabling it to suggest possible meals that could be made from the ingredients. Others have used the software to build functional websites, solely based on a hand drawn prompt for what the website should look like.
This technology has the potential to impact the job market in a fundamentally different way from anything we’ve seen previously. In the past, periods of automation involved the introduction of physical labour-saving devices on the factory floor. In 1970, manufacturing had a 32% share of the British economy: by 2010, this was down to 12%. Undoubtedly, this decline may have partly been caused by problems surrounding Britain’s economic competitiveness, but the changes were also driven by automation.
In response, British governments prioritised sending people to university. Tony Blair set the now infamous target for 50% of young adults to progress to higher education, with the idea that degrees would be able to better equip the country to deal with the rapidly changing economy. With AI, however, it seems that the skills offered by those degrees might be the most under threat. After all, GPT-4 performs exceptionally well in exams. It scored in the 90th percentile of the Bar exam, the 88th in the LSAT, and the 89th in the Maths SAT (I refuse to write the word math). The software has all the cognitive skills that are supposed to make university students competitive in the job market, and it can deploy these skills independently. An analysis by Goldman Sachs estimates that a quarter of current work tasks in the US could be automated by AI, with particularly high exposure in administrative and legal professions, at 46% and 44% of tasks respectively.
This paints a pretty bleak picture for university students; we’ve bought into a view of the economy that might completely break down when we actually enter the workforce.
By contrast, it is the jobs that universities typically don’t serve to prepare their students for, and the ones that were most affected by previous waves of automation, that seem to be safe. Construction, for instance, faces AI exposure in only 6% of tasks. It’s easy to see how AI could analyse legal documents, but it’s not going to come and build you a house anytime soon.
This general view of AI might be a bit bleak, and OpenAI’s software definitely isn’t perfect. In a post-collections essay crisis, I was hoping ChatGPT might take care of this article for me. When asking for the word count of a piece I got it to write, I first got an answer of 390 words (it was actually 537). Further dialogue prompted ChatGPT to keep changing its answer, before deciding that the article was in fact 11 words. It was sure of this. Suffice it to say that counting is not its strong suit.
And even if ChatGPT excels at a task, there are some things that we just want humans to do for us, no matter how good the digital alternative is. OpenAI’s CEO has used chess as an example. Since IBM’s DeepBlue beat Gary Kasparov, computers have been better than humans at chess, yet it’s more popular than ever. And nobody is tuning in to watch two computers go at it on a chess board: our fascination with humans doing that has survived automation.
We also remain integral for displaying creative talent. Youtuber Casey Neistat recently set out to make a vlog written and directed entirely by GPT-4. The result might just be the worst piece of filmmaking I’ve ever seen. Neistat diagnosed this as being GPT-4’s lacking a “soul” in its creative expression. A part of consuming great art is that it allows us to connect with the creator, and the emotions they were feeling as they made it.
AI is clearly going to get better at many things. GPT-4 is already 40% more likely to produce accurate responses than its predecessor. But sometimes it just can’t replicate the impact of a real person.
Sadly, we can’t all become chess players and artists. The majority of us will be forced to reckon with the emergence of a new economy, where the skills that we thought set us apart can now be done cheaply and more effectively by a machine. This might require a societal change in direction. The UK economy is chronically short of skilled tradespeople. It seems those jobs will be golden in a world of AI-based automation. A shift towards training for them would pose a serious threat to the years of high rates of attendance at British universities. But if those institutions are no longer able to prove that they can properly equip their students for the world economy, that might not be a bad thing.
Like the Luddites, there does not seem to be anyone coming to save us from our plight. Of course, in their case, Britain’s industrial revolution was able to make jobs as well as take them away, offering a consolation to those whose jobs faced automation. AI will undoubtably create some new job opportunities, but it’s not clear that this will be sufficient to meet the needs of the new army of unemployed. Without a genuine realignment of our economy, we might find ourselves looking for our own Captain Ludd.
Image credit: Tada Images at www.shutterstock.com
Image description: Webpage of ChatGPT, a prototype AI chatbot, is seen on the website of OpenAI, on a smartphone.
Post Views: 1,885
Liked this article? Why not share it?
Related
Your degree might not save you from AI
In 1779 Ned Ludd, a weaver living in Leicester, broke two stocking frames in a fit of rage. The frames, designed to replace traditional hand knitters, had become emblematic of the technological development that was sweeping Britain, making centuries-old practices redundant. Years later, Ludd’s story would be repurposed into the folkloric character of Captain Ludd, inspiring the Luddite rebellions and resulting in the destruction of textile machines, which were seen by workers as threatening their craft.
It is debatable whether Ludd actually existed, but his status is a recognition of the sense of helplessness workers in industrial England felt in the face of rapid technological change. They saw their only source of help as coming from a mythical figure.
Reflecting on recent developments in Artificial Intelligence, I feel a similar sense of helplessness. The success of ChatGPT, an artificial intelligence chatbot produced by OpenAI, has showcased AI’s power and scope. The Chatbot was estimated to have reached 100 million monthly active users in January, just two months after its launch, making it the fastest growing consumer application in history. Trained from data found on the internet, it seeks to understand and respond to written prompts. In this regard, it’s incredibly good: it can craft resumes, solve maths problems, and even help to write code. Its successor GPT-4, able to respond to image prompts, is even better. One user showed a photo of the inside of a fridge to the bot, enabling it to suggest possible meals that could be made from the ingredients. Others have used the software to build functional websites, solely based on a hand drawn prompt for what the website should look like.
This technology has the potential to impact the job market in a fundamentally different way from anything we’ve seen previously. In the past, periods of automation involved the introduction of physical labour-saving devices on the factory floor. In 1970, manufacturing had a 32% share of the British economy: by 2010, this was down to 12%. Undoubtedly, this decline may have partly been caused by problems surrounding Britain’s economic competitiveness, but the changes were also driven by automation.
In response, British governments prioritised sending people to university. Tony Blair set the now infamous target for 50% of young adults to progress to higher education, with the idea that degrees would be able to better equip the country to deal with the rapidly changing economy. With AI, however, it seems that the skills offered by those degrees might be the most under threat. After all, GPT-4 performs exceptionally well in exams. It scored in the 90th percentile of the Bar exam, the 88th in the LSAT, and the 89th in the Maths SAT (I refuse to write the word math). The software has all the cognitive skills that are supposed to make university students competitive in the job market, and it can deploy these skills independently. An analysis by Goldman Sachs estimates that a quarter of current work tasks in the US could be automated by AI, with particularly high exposure in administrative and legal professions, at 46% and 44% of tasks respectively.
This paints a pretty bleak picture for university students; we’ve bought into a view of the economy that might completely break down when we actually enter the workforce.
By contrast, it is the jobs that universities typically don’t serve to prepare their students for, and the ones that were most affected by previous waves of automation, that seem to be safe. Construction, for instance, faces AI exposure in only 6% of tasks. It’s easy to see how AI could analyse legal documents, but it’s not going to come and build you a house anytime soon.
This general view of AI might be a bit bleak, and OpenAI’s software definitely isn’t perfect. In a post-collections essay crisis, I was hoping ChatGPT might take care of this article for me. When asking for the word count of a piece I got it to write, I first got an answer of 390 words (it was actually 537). Further dialogue prompted ChatGPT to keep changing its answer, before deciding that the article was in fact 11 words. It was sure of this. Suffice it to say that counting is not its strong suit.
And even if ChatGPT excels at a task, there are some things that we just want humans to do for us, no matter how good the digital alternative is. OpenAI’s CEO has used chess as an example. Since IBM’s DeepBlue beat Gary Kasparov, computers have been better than humans at chess, yet it’s more popular than ever. And nobody is tuning in to watch two computers go at it on a chess board: our fascination with humans doing that has survived automation.
We also remain integral for displaying creative talent. Youtuber Casey Neistat recently set out to make a vlog written and directed entirely by GPT-4. The result might just be the worst piece of filmmaking I’ve ever seen. Neistat diagnosed this as being GPT-4’s lacking a “soul” in its creative expression. A part of consuming great art is that it allows us to connect with the creator, and the emotions they were feeling as they made it.
AI is clearly going to get better at many things. GPT-4 is already 40% more likely to produce accurate responses than its predecessor. But sometimes it just can’t replicate the impact of a real person.
Sadly, we can’t all become chess players and artists. The majority of us will be forced to reckon with the emergence of a new economy, where the skills that we thought set us apart can now be done cheaply and more effectively by a machine. This might require a societal change in direction. The UK economy is chronically short of skilled tradespeople. It seems those jobs will be golden in a world of AI-based automation. A shift towards training for them would pose a serious threat to the years of high rates of attendance at British universities. But if those institutions are no longer able to prove that they can properly equip their students for the world economy, that might not be a bad thing.
Like the Luddites, there does not seem to be anyone coming to save us from our plight. Of course, in their case, Britain’s industrial revolution was able to make jobs as well as take them away, offering a consolation to those whose jobs faced automation. AI will undoubtably create some new job opportunities, but it’s not clear that this will be sufficient to meet the needs of the new army of unemployed. Without a genuine realignment of our economy, we might find ourselves looking for our own Captain Ludd.
Image credit: Tada Images at www.shutterstock.com
Image description: Webpage of ChatGPT, a prototype AI chatbot, is seen on the website of OpenAI, on a smartphone.
Liked this article? Why not share it?
Related