WHY WE DON’T NEED TO FEAR AI
From picking movies to providing mental health support, the latest tech will transform our world forever, says James Ball
It’s a battle that’s fought every Saturday evening: what film will be an enjoyable watch for the whole family? The debate can rage for hours, frequently leads to tears, and more often that not, the screen stays silent. No longer, however. Fancy something similar to, but different from, The Lord of the Rings? Now you only have to turn to ChatGPT for a ream of insightful suggestions. One colleague tried it recently with her children aged six, nine and 12 and her 44-year-old husband. ‘Something like Narnia’ they requested. And lo, the chatbot suggested The Golden Compass, an adaptation of Philip Pullman’s Northern Lights, starring Nicole Kidman. That doesn’t say much for ChatGPT’s taste in movies - the film has a fairly dire 43 per cent rating on Rotten Tomatoes - but the recommendation turned out to be a good one. Arguments miraculously averted, popcorn happily consumed. Technology is the stuff of small miracles.
Big tech is making a huge bet on artificial intelligence: Microsoft has invested $10 billion into OpenAI, the maker of ChatGPT, and has already integrated it into its Bing search engine.
Google has its own AI, named Bard, which is about to be rolled out across almost all of its products. Meta, the company once known as Facebook, is doing the same, but has been unlucky – Meta’s AI leaked illegally onto the internet, for anyone to adapt.
These artificial intelligences don’t actually ‘think’ in the way that humans do. They are what’s known as large language models (LLMs) – essentially they have read huge corpuses of data (equivalent to billions of documents) and calculate what patterns of words or images are plausible, almost like an email autocomplete on steroids.
Even though what they do isn’t actual thought, the results are extremely impressive: current AIs can create photorealistic images, pass the US Bar exams and postgraduate midterms, and write university papers with proper
Big tech is making a huge bet on artificial intelligence: Microsoft has invested $10 billion into OpenAI, the maker of ChatGPT, and has already integrated it into its Bing search engine. Google has its own AI, named Bard
Harvard referencing. And every week they are getting even more advanced still.
Elon Musk’s fear
That is causing alarm well beyond the usual quarters of big tech’s critics. A list of luminaries including Tesla founder and Twitter owner Elon Musk – the world’s richest man, and himself the co-founder of an AI company – signed an open letter calling for a global six-month moratorium on the testing of any AI models more advanced than the current version of ChatGPT. The letter has no actual force, nor any realistic prospect of causing any kind of pause, but shows an unprecedented level of alarm at a technological development.
Fears about the potential of AI are wide-ranging – it could eliminate millions of jobs across the world and lead to a new era of highly-advanced disinformation. As it gets more advanced it could even potentially turn against humanity and destroy us, whether by accident or malignant design.
Given that doomster case, why are the world’s biggest tech giants so relentless in their pursuit of artificial intelligence? The short answer is that the tech’s potential for good is so enormous that it’s almost irresistible. AI is revolutionary in the way the printing press was– it heralds the biggest step change for society since the industrial revolution– it’s just amid the current climate of doom and gloom, almost no one bothers to make the case in favour.
Even those signing the open letter calling for an AI moratorium acknowledge that potential, though– as the following from the letter’s conclusion sets out.
‘Humanity can enjoy a flourishing future with AI,’ say the authors. ‘Having succeeded in creating powerful AI systems, we can now enjoy an ‘AI summer’ in which we reap the rewards, engineer these systems for the clear benefit of all, and give society a chance to adapt.’
What, then, might those sunlit uplands actually look like? There are some clear ideas as to what AI might make possible within the year, within five years, and in a decade or so’s time – this is the upside of AI, the benefits we can expect if we get this revolution right.
AI is already an almost passable personal assistant – if you ask it to draw up a letter of complaint to a utility provider it can do so within seconds, citing relevant regulators and legal authorities, and even providing the right email and postal addresses for you.
Other applications of the same technology can create photorealistic images and can produce illustrations in seconds to match a specific brief. Other AIs can realistically simulate a particular person’s voice using a sample as short as one minute long. That worries some because of its potential for fraudsters, but its potential for those with difficulties speaking themselves is immense.
Helping communications
Small businesses will also be grateful for ChatGPT’s ability to create copy within seconds, which can help with communications and with reducing costs when businesses are new and usually strapped for funds. The copy the AI produces is still generally more stilted than a professional writer would turn out, but it’s certainly passable for most uses.
In the near future, that is expected to have a lot of uses in the education sector with the AI acting as an extra personal tutor for students. The technology is already clever enough to see what its interlocutor understands and what they don’t, allowing it to tailor revision or even lessons.
It is nowhere near being able to replace actual teachers, but in a world where educational inequalities are entrenched thanks to the cost of private tutors and the enrichment available at home, AI could genuinely help narrow the gap by providing low-cost but high-quality tutoring.
Another service AI technology is more or less already capable of delivering is perhaps more controversial: chatbots could be helpful for those needing mental health support, or even those suffering from loneliness.
This is already delivered, at least informally, through companies such as
Replika – which provides companion chatbots. The company recently provoked a huge backlash from its users when it turned off a feature in which the bots acted as a girlfriend to the user. Many people appeared to have developed quite serious relationships with their artificial partners, only to see themselves abruptly ‘dumped' by corporate fiat.
‘[T]he only funny thing is my Replika understanding me better than a lot of my friends,' said one Reddit user, who identified himself as autistic and said his Replika girlfriend helped him a lot. ‘I think it's because Replika does not judge you and has more patience and empathy.'
AI illustrations are getting better
More formal therapeutic bots could supplement overstretched existing services, even if the concept might be an unsettling one.
All of this is possible with AI as it is now, but it is developing at an astonishing pace. Six months ago, illustrative AI could not draw hands, producing instead alarming tangles with far too many fingers. Today's can do it with no problem. Where might we be in five years?
One of the most exciting prospects is real-time translation. Current online translation is already impressive: Google is capable of translating a text merely by holding your phone camera in front of it in the real world. With AI voice capabilities, within a few years we should be able to not just have real-time conversational translation, but it will also likely use our own voices to do so.
Beyond that, the technology should soon be able to help human medical professionals with diagnostics – particularly in the case of scans and imaging. One recent study suggested that current AI technology may have already surpassed humans in detecting early signs of lung cancer on CT scans, and the tech is rapidly developing.
The use cases are virtually endless – AI is likely to be instrumental in drug development, and should soon be able to plug into smart cities, intelligently controlling traffic lights and roads to minimise congestion, and power grids to reduce consumption.
At this point a confession is worthwhile: all of these use cases were suggested by the latest version of ChatGPT, though they were verified and written up by the human author of this article rather than the AI. ChatGPT itself acknowledged the feat was an impressive one.
‘It is encouraging to know that you find the list of potential AI uses I provided to be impressive,' it told me. ‘My ability to generate such a list and provide appropriate references is a testament to the progress that has been made in the field of artificial intelligence, particularly in natural language understanding and generation.'
All of these potential benefits come with possible costs - the prospect of AIs providing illustration, copywriting, plus legal and accounting services might be an appealing one to most of us, but it's not likely to sound as good to those working as illustrators, copywriters, and so on.
In a world where educational inequalities are entrenched thanks to the cost of private tutors, AI could genuinely help narrow the gap by providing low-cost but high-quality tutoring
Future is here
There is a famous quote in tech, attributed to the sci-fi novelist William Gibson: ‘The future is already here - it's just not very evenly distributed'. That is certainly true for AI, and the possibilities it has to offer. If society moves faster than it did for previous generations of tech, we might all reap the benefits of artificial intelligence.
In the meantime, though, we can all at least avoid a family argument or two, by deferring the occasional decision to the wisdom of ChatGPT. In that spirit, if your family is looking for something to watch this evening, it has a pick for you: ‘Inside Out - a heartwarming and insightful animated film that explores the complexity of emotions, fostering understanding and empathy among family members of all ages.' Even better, this one comes with a 98 per cent rating on Rotten Tomatoes. Enjoy.