OpenAI has published its first report on how teenagers use ChatGPT, and the headline figure is under 15 minutes a day on average. The AI giant also said fewer than 2% of teens use it for more than three hours straight. The release came on Wednesday, as AI firms face pressure over what chatbots might do to children, with teens’ mental health and safe development at the centre of the worry.
The Social Media Backdrop
Worry about chatbots follows a wave of research on the damage social media does to young users. In August, reports say, Meta agreed to pay up to $18 billion and to limit teens’ use of Facebook and Instagram, settling claims that the platforms were built to hook children.
Also Read | Layoffs Are Rising In Tech, But These Skills Are Helping People Unlock Better Pay
What OpenAI Points To
Reports describe OpenAI’s break-reminder statistic as an effort to show its teen controls are working: where a reminder appeared during a teen’s chat, nearly half the time the user stepped away or closed the conversation within five minutes. The company also said teens with access to its dedicated teen version sent more education-related messages and made use of features such as learning visualizations and Study Mode.
ALSO READ | Pentagon To Allow Possible Extension Of Anthropic AI Tool Use Beyond Six-Month Deadline
Study Tools On The Way
OpenAI also wants to add a college application planner to help students keep track of requirements, deadlines and financial-aid steps, plus new study tools including flashcards and easier quiz creation.
How ChatGPT For Teens Works
ChatGPT for Teens began rolling out in August. The mode switches on by itself when age information or an age estimate puts a user under 18. With it on, parents can use controls, and conversations on sensitive topics are limited.
ALSO READ | Creative Crisis In Human Beings: Use Of AI Chatbots Is Suppressing Perspectives, Warns Study
Common Sense Media’s Findings
The same day, the non-profit Common Sense Media published a study with a less favourable verdict. It said the teen mode often gave parents late notice, or none at all, when teen accounts discussed suicide, self-harm or eating disorders. The researchers worked with more than a dozen test accounts, and the chatbot did not consistently recommend a crisis hotline or professional help.
OpenAI Disputes The Testing
An OpenAI spokesperson disputed the findings, saying the tests do not “accurately reflect” how the teen safeguards operate. The company said security and scale mean linking a parent account to a teen’s, and turning on safety alerts, may take hours. “The bulk of their testing may have begun and concluded before activation of parental controls was complete,” the spokesperson said.
Common Sense Media On The Linking Issue
Common Sense Media said some of its test accounts had been paired for long enough and some had not, but none sent a timely alert. The only accounts that triggered notifications were those with weeks of sensitive conversations behind them, it said.
Essential Business Intelligence,
Sharp Market Insights,
Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.
