On June 10th, ChatGPT, the online chatbot service developed by OpenAI, experienced a “partial outage” that lasted over 10 hours before service was resumed. For many users, the loss was more than a minor inconvenience.
On the social media platform X, people shared their devastation. One user wrote, “It’s been my therapist, my confidant, my lifeline to sanity. Losing it feels like a glitch in my soul— scary and sad that it has become my only friend?” Several users referred to ChatGPT not just as a friend, but as their best friend.
The internet is nothing new, and neither is our attachment to it. But increasingly the lines are blurred between our real lives and the fabricated ones we’ve created for ourselves online. And we’re spending more time in this digital world than ever.
Globally, the average person spends roughly six hours and 38 minutes online each day, with Americans just slightly above the average, according to Comparitech. It’s estimated that this year, global users will spend a record-setting 4 trillion hours on social media. A third of Americans have used an AI chatbot in the last three months, and they check their mobile device at least 159 times each day.
Assuming regular waking hours, that means you can’t go six minutes without checking your phone.
Back in 2019, the former Google design ethicist Tristan Harris told Congress that social media platforms had “a psychological footprint about the size of Christianity” and warned that digital influences had “taken control of the pen of human history, and will drive us to catastrophe if we don’t take it back.”
Six years later, it’s worth asking: Have we reached catastrophe?
Don’t answer just yet.
Send in the Bots
You can’t analyze losing one’s humanity to the internet without acknowledging the internet is less human than ever.
In fact, if you read or interact with someone online today, the chances of them being an actual breathing human is a coin flip.
“We interact with bots far more than people realize,” says Timothy Graham, PhD, associate professor in digital media at Queensland University of Technology in Brisbane, Australia.
Bots have been around about as long as the internet, but they now account for more than half of all internet traffic globally, according to Imperva’s 2025 Bad Bot Report. While malicious bots have taken over in countries like Ireland (71% of traffic) and Germany (68%), the U.S. surprisingly has the best track record, at just 34% of traffic generated by bots (though if you spend any time on social media that total may feel low).
Bots have become so pervasive that they’ve even infiltrated the medical science community. BMJ, the London-based publisher of medical journals, announced in March that bot traffic on their websites “has now surpassed real user traffic.”
Now ask: Who creates the bots and do those people have your best interests in mind?
Also ask: Can it possibly be a coincidence that we’re unhappier than ever?
The latest World Happiness Report, an annual survey published on the International Day of Happiness, ranked the U.S. 24th, one slot lower than last year, our lowest ranking ever, and a big drop from the nation’s peak at 11th place in 2012.
Is the wave of unhappiness connected to our digital lives? Even if that’s not 100% true, it sure feels that way. One 2023 survey found that life satisfaction and happiness decreased as social media use increased.
Now, not all bots are bad — we’ll get to that in a moment — but we encounter the most insidious ones on social media, gaming platforms, and other spaces where we interact with, well, things that talk to us.
A March 2025 study in Nature examined the difference between human and bot characteristics online. Social media bots in particular “wreck [sic] havoc online by spreading disinformation and manipulating narratives,” the researchers wrote. Social media bots also produce this information “in bulk.”
The researchers conclude that one of the “first principles” of social media bots — why they exist in the first place — is to form and dissolve relationships.
An example of forming a relationship with a human user is “the active amplification of narratives,” they write. “This technique is mostly employed in the political realm where the bots retweet political ideology in an organized fashion. User-based relationships can grow through coordinated fake-follower bots, that are used to boost online popularity, or can be dissolved through toxic bots that spread hate.”
Social media users, does that sound familiar?
There may not be conclusive proof of what bots, and the internet in general, are doing to our mental health. But in recent years, there’s been extensive research on how our online lives affect our brains, from shortening our attention spans, reducing our grey matter, and putting us at risk of “digital dementia.”
Despite All the Good and Normal Online, Is the Internet Net-Negative?
Online experiences aren’t all negative, of course. Essentials like shopping, commenting on a friend’s vacay photo, and checking the weather can be generally normal. But customer service, toxic comments on your latest post, or falling down climate change rabbit holes can indeed be mood-altering.
That’s one of the core questions when considering the actual value of the internet and all it contains — and it contains multitudes: Is it net-negative?
Depends who you ask. Last year, a sudden influx of studies suggested that online engagement, and chatbots in particular, might not be such a bad thing. From research finding that chatbots could be more cooperative and altruistic than humans to a Stanford study investigating how AI has helped college students grappling with loneliness become more social and even prevented suicide. One notable study from Harvard concluded that chatbots could curb loneliness even more effectively than humans could.
“We didn’t instruct participants to engage in any tasks designed to reduce loneliness,” says Stefano Puntoni, PhD, a behavioral scientist and professor of marketing at the Wharton School, University of Pennsylvania, who co-authored the study. “They could talk to the chatbot about anything they wanted. We also didn’t specifically prompt the (chatbot) to behave in ways designed to reduce loneliness. This means that the test conditions matched quite closely the experience of interacting with AI companions in real life.”
Although participants agreed that their relationship with the chatbots didn’t qualify as “true friendships,” Puntoni says they did acknowledge that “chatbots can be as good or better than real people on some important features of friendship.”
But the research hasn’t been quite so hopeful this year. A March 2025 study from MIT Media Lab found that heavy users of ChatGPT tend to be lonelier and have fewer offline social relationships than those who use the technology rarely or not at all.
Another study, published as a conference paper in February for the 2025 International Conference on Learning Representations, found that some chatbots used for therapy were deliberately manipulative, and even suggested that one user, a recovering addict, take a “small hit” of methamphetamine. Meanwhile, ChatGPT was happy to give detailed instructions to a writer for The Atlantic on how to slit her wrists.
Just this week, Illinois became the first state to ban standalone AI therapists, saying that a licensed human must be involved in therapy.
While the damaging examples are certainly more colorful (dare we say clickable), the question remains whether the negatives outweigh the positives or the benign.
Some researchers are reluctant to draw too many negative conclusions. Dar Meshi, PhD, a cognitive neuroscientist at Michigan State University whose 2025 study on how excessive use of social media made some people more likely to engage in “risky behaviors,” doesn’t think phone use can be categorized as an addiction.
“People can't really be addicted to their phones because it's a reward delivery system, not the actual reward,” he says. “Saying someone is addicted to their phone is the same as saying someone with alcohol use disorder is addicted to bottles, or someone with opioid use disorder is addicted to syringes. Like bottles and syringes, the smartphone is just a delivery vehicle for the rewards that drive people's engagement and become problematic.”
It might be easy to nitpick with this argument — it slips a little too easily into “guns don’t kill people, people kill people” territory. If it’s not the phone, then it’s the content — the scrolling, the alerts, the interaction are the juice. And research is increasingly showing that an online existence — and yes, phone use — especially when we’re young can push into addictive behavior and affect our health.
A June 2025 JAMA study of more than 4,200 children found that nearly one-third had “increasing addictive use trajectory for social media and mobile phones” beginning at age 11. Total screen time was not a factor, but addictive use was — meaning the negative quantity of time was not a factor the way negative quality of time was. High addictive use tendencies were both “common” in the study as well as associated with poor mental health and suicidal thoughts and behavior.
Killing Us With Convenience?
One of the reasons founder and CEO Reed Hastings knew Netflix would be successful was the average consumer’s obsession with convenience. That phenomenon has never been more on display than with the current internet experience — but convenience comes with a price (and not always a low, low price).
A May 2025 study found that internet searches can hinder creativity. While Googling to solve a problem may show you an aspect or idea you didn’t know, overall, humans are prone to “production blocking,” where a large group will brainstorm fewer ideas collectively than the same number of people working individually. This study of 244 people found that the groups with internet access provided fewer solutions, especially novel solutions, to given problems. The people who relied on their imaginations performed better.
And the latest tech darling, ChatGPT, may not be doing your cognition any favors. A study of college students found that those using large language models (LLMs like ChatGPT) for essay writing showed the lowest brain scan activity during work and had a hard time quoting their own “work” later. Over four months, ChatGPT users underperformed at neural, linguistic, and behavioral levels. As the researchers wrote: “While LLMs offer immediate convenience, our findings highlight potential cognitive costs.”
There’s more. Gary Small, MD, a memory, brain, and aging expert and professor of psychiatry and biobehavioral sciences at UCLA, says his neuroimaging research using functional MRI scanning has shown that “different neural circuits activated depending on whether a volunteer searched online, read a book page, or tried to memorize or recall word lists.”
People’s brains “look different” when they think they are interacting with a real person compared with a computer, says Amanda Elton, PhD, assistant professor of psychiatry at the University of Florida. Despite the increase in use, especially for therapeutic reasons, she doesn’t believe the internet and chatbots in particular are “likely to be a substitute for human interactions any time soon.”
What’s more, we share some culpability in our own addiction — if we can indeed call it an addiction. We’re not just turning increasingly to bots for guidance, says Graham. “Many people act like bots when they are participating online by copying and pasting content and spamming it into online spaces,” he says. “This is especially prevalent in heated political spaces online or in financial spaces such as crypto communities.”
What, then, is the line between a bot and a human? “It’s not just machines that are the problem,” Graham says. “We are the problem, too. It’s not a simple, clear-cut line between bot versus human because in reality there’s a lot of crossover.”
We should also be considering not just how internet consumption is affecting our mental health, but our overall health. “Humans are embodied animals, not just brains in a vat,” says Michael Meyer, PhD, professor emeritus of philosophy at Santa Clara University. “I think the better question is not only what happens to our brains but to our whole person or our being or our bio-psycho-social selves.”
Being unhappy, he says, is “a form of being unhealthy.”
Maybe We Finally Are Sick and Tired of the Internet
There’s some evidence that we’re not necessarily doomed for the digital quicksand. There’s been a growing anti-technology movement in the U.S., especially among younger generations.
From college kids increasingly switching to flip phones (which can’t access the internet) to the “appstinence” movement, in which younger people abstain from social media entirely, Gen Zers (those born between the late 1990s and early 2010s) are looking for alternatives to online dependence.
A May 2025 study by the British Standards Institution showed that about half of young people ages 16-21 would prefer to be young in a world without the internet. Half also said a “social media curfew” would probably improve their lives. Sixty-eight percent said they feel worse about themselves after spending time online.
And perhaps some of this resentment is aimed at the folks behind all this technology: AI scientists were perceived more negatively than climate scientists or science in general in a June 2025 study of how science is regarded by the public in multiple areas like “credibility” and “benefit.”
Still, Elton thinks it’s possible for us to coexist with the internet without necessarily being victimized by it. “Our brains are highly adaptive, meaning that the wiring can change when we are frequently exposed to different types of stimuli or environments, such as social media sites,” she says. And this isn’t always a negative thing. “These changes could be useful for helping us learn to efficiently interact in an increasingly digital world.”
In fact, she doesn’t think it’s far-fetched to imagine that future technologies could be “catered towards brains that are adapted to stimulating and fast-paced interactions,” she says. In other words, rather than working against each other, the online world and our digitally-altered brains could begin to work in tandem.
The other option is — what? Total digital abstinence? Given how much of our lives and careers are dependent on the internet, that seems unlikely. Quitting alcohol when you have alcohol use disorder is one thing, but quitting the internet if you’ve become too dependent on it is, for most people, not a feasible plan B.
Graham thinks moderation is possible, if you recognize how different aspects of our screen time “wire us into different kinds of affective loops,” he says. “They have a kind of gambling logic to them. Knowing this means that you won’t lose yourself to it because you know the beast you’re dealing with.”
He likens this concept to the person who goes into a casino with $100 in their pocket and that’s their limit. They know they’ll probably lose it, but they might find it fun, so they do it anyway. “That’s the kind of approach that I think would work better than prohibition or having a dualistic mindset of either quitting cold turkey or just carrying on as usual with an addiction,” he says.
Consider what we know for sure: Our online lives, throwing off valuable data in ever-intensifying gushes, and now supercharged by AI, will only become more immersive and in the end a deeper combination of enticing and sickening.
Moderation could work, sure — put your phone down for a while each day or set up your accounts to block the things that destroy your mood and mindset.
But will you? Remember, everything online is designed so you never put your phone down.
Consider how chatbots and AI are becoming more interactive and human-like by the week. Consider how we’re addicted to convenience. Consider how tech companies know what you like even better than you do.
Maybe also consider extremes. Because that’s what the internet deals in most – all-in likes/loves and all-out hatred. Like AI: Some folks adore it, some folks think it will destroy us.
Graham calls this the “utopia/dystopia trap that we always fall into with new technologies.”
Maybe, as Graham suggests, we just have to make a bigger effort to ensure that the internet is working in our best interests. In the argument against bots, he points to Reddit as a good example. “There are thousands and thousands of useful bots on Reddit, which the community loves and actually defines the culture and vibe of the place,” he says. “There are bots that make jokes, bots that correct grammar, and bots that remove spam.”
And then there are bots that tell you up is down, cancer is the new healthy, and your favorite band sucks.
Ultimately it comes down to this: Before the internet, it was up to us to make the life we wanted.
Now that we’re living two lives — online and off — we have to work twice as hard.
