AI Overdependence The Unseen Risks Of ChatGPT

It’s a normal Saturday afternoon in central London, and my friend and I are deciding where to have lunch. I squint around at the nearby cafes, and reel off some of the places I’ve been to before. My friend, however, is occupied with punching something into her phone. “I just asked ChatGPT,” she declares, listing off the bullet-pointed pros and cons of every single eatery that serves “Italian-style sandwiches” in a 100-metre radius. She is pleased with her findings. But I am disappointed, if not a little mortified, that her default move in that extremely low-stakes situation was to ask an AI chatbot. She could have looked on Google, glanced at a Time Out listing, or, I don’t know, used her knowledge of this city (where she’s lived for the past seven years, I might add).

The AI Crutch A New Norm

It doesn’t sit well with me. Call me a luddite, but I find myself increasingly concerned about how reliant those around me are becoming on artificial intelligence, particularly OpenAI’s popular generative language model ChatGPT. My closest friends and my partner use the chatbot as an informational crutch on a daily basis, swearing off the humble Google search for ChatGPT’s instantaneous, personalised responses. Some friends use it for mundane queries; my boyfriend uses it as his digital assistant (which he affectionately calls “Chat”). Others take it further: I’ve read worrying stories of people using ChatGPT as their therapist, to help craft responses to their spouse during an argument and students relying on it for school or university assignments as institutions continue to grapple with this new advanced technology.

Balancing Benefits with Rising Anxieties

Don’t get me wrong: I understand the benefits of using AI. I’ve heard success stories of people who use it to create workout plans, generate images for product research, or use it as a personal assistant to help kickstart a business. You’ve got all of the answers you need at your fingertips (even though accuracy is not guaranteed, and it can, frequently, be wildly wrong). But with mounting concerns about how the advancement of AI will impact the future of work, the climate and our declining cognitive skills, I am embarrassed by how chatbots are making us downright passive. It’s as though nobody has a desire to problem-solve any more.

ChatGPT, the most popular software for this activity with 500 million weekly users, is a type of generative AI, which means that it creates new content from massive datasets based on publicly available sources on the internet. If you were to search “the best place to get coffee near me” on ChatGPT, for example, you would receive detailed results with pretty pictures, menu options and opening times. A Google search, by comparison, would require trawling through a couple of sites and drawing your own conclusion. That may seem like a very convenient and harmless tool, but when you consider the moral conflict surrounding AI usage, such as the fact that US data centres powering AI produce carbon emissions that rival the domestic aviation industry, it has become the latest social conundrum we must navigate within our relationships.

The Personal Divide AI Adoption and Its Social Ripples

Laura*, a 27-year-old journalist who is staunchly against AI, finds that the topic generates friction when it comes to dating. Her main issues with software like ChatGPT are the environmental impact (one ChatGPT query uses 10 times more energy than a Google search, but more on that later) and that the collective overreliance on it is “making people lazy”. “When I hear people say that they use it for making shopping lists or holiday itineraries, I think, ‘God, use your brain,’” says Laura. If she’s dating someone who uses AI programs regularly, she will set firm boundaries in place. “If someone tells me they use ChatGPT, it’s not a red flag, but it’s an ick. Ideally, I don’t want to date someone who uses it a lot. But when I have dated people who use it, I’ve made it very clear... do not use it around me, and please don’t send me anything you’ve put through it. Normally, they’ve been pretty respectful of that, including people who use it a lot.”

Laura is also aware that she isn’t on the moral high ground. “As much as it’s an ick for me, I’m also aware there are probably people who find it weird that I don’t use it. There’s a lot of judgment on both sides of it, and I’m OK with being judged for not using it.”

As a personal principle, Laura will stop using an online service if she finds out that it has AI integrated within its features, and will actively avoid using it wherever she can. “Personally, it’s not something that I ever want to use or engage,” she says. “I think it is incredibly disappointing how quickly we’ve all come to accept it and be like, ‘Oh well.’”

In April, everyone from celebrities to pro sports teams hopped on the latest AI-generated social media trend, using ChatGPT to create an image depicting themselves as action figure, cased inside Barbie-like plastic packaging. One friend of mine generated a Bratz-style figurine of herself, which held her favourite coral-coloured Stanley cup in her right hand; her mobile phone in her left. The trend received backlash, mostly from artists and creatives who argued that AI art isn’t art – and that partaking in this trend was risking their jobs. Laura was similarly alarmed by the number of people she knew taking part, too. “I saw influencers, who claim to be vaguely ethical, all jumping on trend, and then soon after, there was another trend where you use an AI voice to narrate a movie trailer for your life. When I see that, I just think it’s so lazy.”

Environmental Alarms The Hidden Ecological Cost of AI

You may argue that AI is making people stupid, but the environmental implications are far, far more alarming. According to a 2024 report released by Climate Action Against Disinformation, the burgeoning electricity demands of AI will require a doubling of data centres, which store information crucial for AI, cloud computing and digital public services. This, the report states, will cause an 80 per cent increase in planet-heating emissions, even if there are measures to improve the energy efficiency of these centres.

Expert Insights Unpacking AIs Energy Appetite

Lucy Atkinson, an associate professor at the University of Texas at Austin who studies the public perception of the environmental impact of AI, tells me that she’s observed how there’s a lack of awareness around how AI is linked to carbon emissions. She explains that data centres require vast amounts of energy, which makes them hot – and therefore lots of water is required to cool them down. “ChatGPT certainly has a larger carbon footprint than other ways that we may use technology,” says Atkinson, adding that key data points she has used in her research are that ChatGPT uses 10 times more energy than a Google search and that 15 queries consume the same energy as watching an hour of video content. “Those statistics may sound modest, but there are more than a billion queries on ChatGPT every day, so that adds up.”

Atkinson’s research is based in Texas, the state with the third most data centres in the US, at approximately 350 statewide. She’s found that through running surveys among Texans, most people don’t realise they live near a data centre, and how the environmental impact of AI could be closer to home. “Data centres are going to need twice as much power in 2027 as they did in 2022 – we are just feeding that beast with queries,” she says. “In the public perception, there’s a disconnect between AI data centres and the environment, when they have this profound impact on the environment.”

As part of her current research, Atkinson asked respondents in Texas if they were aware they lived near a data centre, and two-thirds either said no, or that they weren’t sure. “There’s a very good chance they do live near a data centre – they’re just not aware of it,” she explains. “The water use is the biggest environmental impact, along with the energy required to make them run. Then other issues affect people who live near them, such as noise pollution, increased traffic and other land use changes.” According to the data centre map, there are 428 data centres in the UK in total, and that number will only grow as the demand for AI increases.

The Pragmatists View AI as a Workplace Tool

Jacob Rankin, a 29-year-old graphic designer for an IT company based in London, uses ChatGPT every day at work for productivity tasks such as summarising meetings, sending emails and managing his workload. AI is a hot topic among his friends, and he finds himself having to play devil’s advocate. “A lot of them are left wing and also environmentally conscious, and they bring up those facts about how ChatGPT uses a certain amount of water. The way I see that argument is that it’s always going to be a challenge to get people to care about the climate in those moments when it’s 5pm on a Friday and your boss is like, ‘Can we do this by tomorrow?’ Are you more likely to press the AI button and do it in two seconds, or are you going to do it yourself? The value they have created is a lot higher than the environmental factors in that moment.”

Rankin finds that most scepticism about AI comes from those who aren’t familiar with it. At work, he’s actively encouraged to use AI in his day-to-day tasks – and he’s not worried about it taking his job in the future. “Ultimately, you know where it works and it doesn’t. When the conversations about AI started happening properly a few years ago, I knew I had to be quick to learn about it so that I could sit in a room and tell someone why I might be better than AI. If you work with it every day, you know that you haven’t got anything to worry about.”

Navigating Moral Mazes and AIs Imperfections

Many people I speak to for this article admit some level of moral contradiction around their AI usage, but see it as unavoidable as more companies such as Apple, Microsoft and Meta integrate compulsory AI features into their platforms.

Atkinson herself admits she is “guilty” of using ChatGPT to do things she could probably do on her own. She recently used the chatbot to plan a trip abroad, including finding her flights and creating an itinerary. But that hasn’t gone without its factual errors and inaccuracies. “ChatGPT suggested a flight that I knew didn’t exist anymore. And so I said, Are you sure about this flight? I was being very polite, because that’s what we’re supposed to do with the robots. And it replied very politely, ‘I’m sorry, this is based on old data.’”

Whether you have nightmares about AI or proudly embrace it, there must be a wider awareness that the data it’s pulling from could be wrong, and we should stay questioning the results that it is churning out. “Sometimes I ask ChatGPT to produce images for me, and those are of questionable quality,” says Atkinson. “Although it gets better every day.”

The Human Edge Why Critical Thinking Still Matters

ChatGPT may have hijacked my friends and family, but it’s far from being as capable as a real, thinking human, so we can take solace in that. At least for now.

Names have been changed