Trending Articles
Trending Topics
Trending
Meticulous. Delve. Adept. There’s nothing wrong with these words, but their usage in video speech shot up significantly after the initial release of ChatGPT.
That’s according to researchers at Germany’s Max Planck Institute for Human Development, who studied 280,000 YouTube videos to spot the impact of OpenAI’s LLM-powered bot. And isn’t just language: research from MIT and Microsoft have suggested that using generative AI impacts our cognitive skills.
Put another way, we’re melting our brains. That matters as AI increasingly creeps into workplaces, schools and even home use.
To unpick the impact of AI bots on our speech, the Max Planck researchers transcribed and analysed 280,000 English-language videos from 20,000 YouTube channels associated with academic institutions – in other words, not your average YouTuber, but a collection of presentations, talks and speeches – to analyse changes in word frequency before and after ChatGPT.
According to a preprint article, they found increases between 35% and 51% in the words delve, realm, meticulous and adept. These words were all chosen for study as they were identified as commonly used in ChatGPT generated content.
More generally, the academics spotted an increase in the use of prowess, tapestry and intricate, with a decrease in groundbreaking, unearth and nuance, among others.
As the researchers studied academic presentations and speeches, this doesn’t necessarily suggest such words are slipping into everyday speech. It more likely indicates that people are turning to ChatGPT and other AI tools to write such material before delivering it verbally.
The researchers added that they “anticipate that similar patterns may extend to other communicative contexts,” before adding that most hackneyed of phrases that more research is required.
The study notes: “Our results raise societal and policy-relevant concerns about the potential of AI to unintentionally reduce linguistic diversity, or to be deliberately misused for mass manipulation. They also highlight the need for further investigation into the feedback loops between machine behavior and human culture.”
We may sound smarter by expanding our day-to-day vocabularies to include “delve” and “meticulous” more frequently, but AI may well be making us less thoughtful.
In a separate study from MIT’s famous Media Lab, researchers asked a set of students to write essays, giving one group access to ChatGPT, a second group the ability to use Google Search, and the third no tech assistance.
The academics then used an EEG to measure the subjects’ brain activity as they worked on the task, finding significantly different neural connectivity patterns across each.
“Brain connectivity systematically scaled down with the amount of external support: the Brain‑only group exhibited the strongest, widest‑ranging networks, Search Engine group showed intermediate engagement, and LLM assistance elicited the weakest overall coupling,” the paper notes, adding that the LLM group were the least likely to be able to recall what they’d written in their essays only minutes later.
That raises serious concerns given the rise in AI use in schools – the output might be deemed high enough quality to score a solid grade, but students aren’t benefiting in the long run.
That isn’t just a problem for students: a Microsoft/Carnegie Mellon study from earlier this year showed use of AI at work impacted critical engagement, in particular for lower-stakes tasks. When users lacked trust in the results, or the task at hand was important, they were more likely to double check the output, and therefore engage their own critical thinking capabilities.
More research is needed on these subjects – of course – but AI is moving into schools and workplaces faster than academics can run these studies.
So perhaps it makes sense to proceed with caution in how we use AI, lest we find ourselves no longer adept enough at critical thinking to delve into such meticulous research in the future.
