
ChatGPT is getting smarter, but excessive use could destroy our brains, study warns
Is it an artificial lack of intelligence?
Not only is AI getting frighteningly smart, but it may be making us dumber as well.
Scientists found that students who used ChatGPT to complete essays had poorer cognitive skills than those who relied on just their brain, according to a dystopian new study out of the Massachusetts Institute of Technology (MIT) in Cambridge.
Advertisement
'Reliance on AI systems can lead to a passive approach and diminished activation of critical thinking skills when the person later performs tasks alone,' the researchers wrote, per the Telegraph.
The team had set out to determine the 'cognitive cost' of using large language models (LLMs), which have become increasingly omnipresent in every sector of society, including academia. According to a winter survey by the Pew Research Center, approximately 26% of teen students used the AI chatbot to help them with assignments in 2024 — up from just 13% in 2023.
3 ChatGPT is easily accessible on smartphones and other tech.
Ascannio – stock.adobe.com
Advertisement
To determine how using synthetic homework assistants affects the mind, the MIT researchers tasked 54 people with writing several SAT essays, Time Magazine reported. Participants were split into three groups: one that relied on pure brainpower, one that used Google, and a third that enlisted the aid of the now-ubiquitous LLM ChatGPT.
Each person was outfitted with an electroencephalography (EEG) device so researchers could monitor their brain activity while completing the task.
They found that the ChatGPT group 'performed worse than their counterparts in the brain-only group at all levels: neural, linguistic, scoring,' according to the Telegraph.
3 'Reliance on AI systems can lead to a passive approach and diminished activation of critical thinking skills when the person later performs tasks alone,' the researchers wrote.
PhotoGranary – stock.adobe.com
Advertisement
The readings also showed reduced activity in the regions of the brain associated with memory and learning, the authors said, noting that a lot of the 'thinking and planning was offloaded.'
In fact, AI-aided scholars got lazier with each subsequent paper to the point that by the third essay, they were simply typing the prompt into ChatGPT and having it do all the work.
'It was more like, 'Just give me the essay, refine this sentence, edit it, and I'm done,'' said the paper's main author, Nataliya Kosmyna.
By contrast, the essayists with no external aid demonstrated the highest levels of neural connectivity, especially in regions of the brain responsible for language comprehension, creativity and memory.
Advertisement
The brain-only group was also more engaged and satisfied with their essays, per the study.
Interestingly, the Google group showed just slightly lower levels of engagement, but the same amount of recall — a perhaps troubling prospect given the increasing number of people who dive into research using AI rather than internet search engines.
3 Researchers deduced that too much reliance on AI could have long-term cognitive effects.
Daniel CHETRONI – stock.adobe.com
Researchers deduced that 'frequent AI tool users often bypass deeper engagement with material, leading to 'skill atrophy' in tasks like brainstorming and problem-solving.'
That could have long-term ramifications, including 'diminished critical inquiry, increased vulnerability to manipulation' and 'decreased creativity,' the authors said.
Fortunately, the findings weren't a total indictment of AI in academia.
As a follow-up exam, the scientists asked the ChatGPT group and their brain-only counterparts to rewrite one of their previous essays — but the AI-assisted participants did so without the chatbot, while the unassisted group could use the cutting-edge tech.
Advertisement
Unsurprisingly, the original ChatGPT group didn't recall much info from their papers, indicating either a lack of engagement or an inability to remember it.
Meanwhile, the former brain-only group exhibited a marked increase in brain activity across all the aforementioned regions despite using the tool.
That suggests if used properly, AI could be a helpful academic tool rather than a cognition-destroying crutch.
Advertisement
The warning about AI-induced brain atrophy comes — somewhat frighteningly — as the technology is becoming more 'intelligent.'
Recently, Chinese researchers found the first-ever evidence that AI models like ChatGPT process information similarly to the human mind — particularly when it comes to language grouping.

Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Yahoo
2 hours ago
- Yahoo
16 billion passwords from Apple, Facebook, Google and more leaked. Why has no one heard of it?
Sixteen billion passwords to Apple, Facebook, Google, and other social media accounts, as well as government services, were leaked in what researchers are calling the largest data breach ever, according to reports. The leak exposed 16 billion login credentials and passwords, prompting both Google to tell billions of users to change their passwords and the FBI to warn Americans against opening suspicious links in SMS messages, according to a report published Thursday in Forbes. Researchers at Cybernews, who have been investigating the leak, found '30 exposed datasets containing from tens of millions to over 3.5 billion records each.' All but one of these datasets have not been previously reported as being exposed, so the data impacted is all considered new. 'This is not just a leak – it's a blueprint for mass exploitation,' the researchers said. And they are right. These credentials are ground zero for phishing attacks and account takeover. 'These aren't just old breaches being recycled,' they warned, 'this is fresh, weaponizable intelligence at scale.' Most of that intelligence was in the format of a URL, followed by logins and passwords. That information then allowed access to 'pretty much any online service imaginable, from Apple, Facebook, and Google, to GitHub, Telegram, and various government services.' While worrisome, the researchers found that the datasets were exposed very briefly – with enough time for them to be discovered, but not long enough for researchers to figure out who was controlling the data. Researchers have determined the leak is the work of multiple infostealers, but it's impossible to tell how many people or accounts were exposed, according to Cybernews. The experts urge people to invest in password management solutions, not share their passwords and to stay alert in the event their passwords are compromised. Error in retrieving data Sign in to access your portfolio Error in retrieving data Error in retrieving data Error in retrieving data Error in retrieving data


Android Authority
2 hours ago
- Android Authority
Will Android 'App Cast' let you stream apps from device to device? (APK teardown)
Rita El Khoury / Android Authority TL;DR Google Play Services shows evidence of an in-development 'App Cast' tool. While its full functionality is unclear, it appears able to gather info about remote Android devices and access their installed apps. One possibility could be the ability to remotely stream apps from one Android device to another. Google's always working on something new for Android, and we spend a fair amount of time digging through app and system updates in the hope of getting an early look at some of those developments. When we're lucky, evidence is abundant and we're able to easily infer what Google's up to. Other times, however, it's a bit harder to get the full picture of what Google might be working on — even when it seems clear that developers are definitely up to something. And today, we're finding ourselves much more in the latter camp. Authority Insights story on Android Authority. Discover You're reading anstory on Android Authority. Discover Authority Insights for more exclusive reports, app teardowns, leaks, and in-depth tech coverage you won't find anywhere else. An APK teardown helps predict features that may arrive on a service in the future based on work-in-progress code. However, it is possible that such predicted features may not make it to a public release. We're looking at a recent update to Google Play Services, and as you may know, this software is how Google implements a lot of system-level functionality in Android — like we looked at just last month when we spotted new device backup settings. This time we've been crawling through the 25.24.31 beta release, and have spotted evidence of a new 'App Cast' feature. We see references to App Cast popping up in code related to multi-device sync. We've also identified a number of text strings that appear to have been added in preparation of App Cast support: Code Copy Text Starting app cast for package %s Fetching app list from remote device. Unable to get icon: %s not found Starting cast session for ' ' from ' Launching '%s' on remote device '%s' We haven't been able to surface a UI for App Cast, and so we can't see any of this in action, but one possibility could involve Google giving Android devices the ability to stream apps installed on other hardware linked to your account. We can tell that there are elements in place for choosing devices from a list, and showing a progress bar. And as mentioned in the strings above, we see code for pulling up a list of installed apps. Admittedly, the need for an app streaming mode feels a little narrow — in a lot of cases, it seems easier to just install the app on multiple devices — but we can appreciate times that require access to data that lives on a specific phone or tablet. We also wonder if this casting might extend to bigger screens, letting you view Android apps from your phone on Google TV. But as we said, right now we just don't have enough of the puzzle to really say with any kind of certainty what Google is building here; this is all vibes at the moment. We've also spotted work in Play Services on some manner of new notification sync support that appears in proximity to some of this App Cast stuff, but we're not sure if that's just a coincidence, or if they're really meant to be related. That's a little frustrating, but we'll keep digging into future Play Services builds in the hopes that we can soon nail down Google's intentions for App Cast. Got a tip? Talk to us! Email our staff at Email our staff at news@ . You can stay anonymous or get credit for the info, it's your choice.

Engadget
3 hours ago
- Engadget
Midjourney adds AI video generation
AI company Midjourney has released its first video model. This initial take on AI-generated video will allow users to animate their images, either ones made in Midjourney or uploaded from a different source. The initial results will be five-second clips that a user can opt to extend by four seconds up to four times. Videos can be generated on web only for now and require at least a $10 a month subscription to access. Introducing our V1 Video Model. It's fun, easy, and beautiful. Available at 10$/month, it's the first video model for *everyone* and it's available now. — Midjourney (@midjourney) June 18, 2025 Midjourney was one of the early names in the space for AI-generated still images, even as other platforms have pushed the forefront of the discussions around artificial intelligence development. Google's latest I/O conference included several new tools for AI generated video, such as the text-to-video Veo 3 model and a tool for filmmakers called Flow . OpenAI's Sora, which debuted last year, is also a text-to-video option, while the more recent Firefly Video Model from Adobe can create video from a text or image prompt. But being a little late to the video game hasn't stopped it from drawing the ire of creatives who allege that its models were trained illegally. In fact, this video announcement follows hot on the heels of a lawsuit against the company. Disney and NBCUniversal sued Midjourney last week on claims of copyright infringement. And as with any AI tool, there's always a potential for misuse. But Midjourney has nicely asked that people "please use these technologies responsibly" so surely nothing will go wrong.