Tech Newsletter

The reddit user going by the name ibreakphotos (how appropriate) has accused Samsung of faking their space zoom feature on their S20-23 Ultra phones. Ibreakphotos believes that the moon photos taken by the phone are just an overlaid image of a high definition moon shot with a better camera. The user took a high definition moon photo and downsized it and also applied a blur to the moon. He then stepped away and took a picture. Sure enough, the phone added more detail than the source image. Whether this is faking the moon or if it is just adding detail to an otherwise terrible photo, is for you to decide. Phone cameras would be terrible if they did not process the images that they take. If you would rather have a blurry mess instead of the moon, you can turn off Samsung’s scene optimizer setting. 

Cerebral, the online therapy startup, admits to sharing patient data with social media giants like Meta, Google, and TikTok. This data included names, phone numbers, IP addresses, insurance information, and much more. Cerebral claims that this sharing of data was unintentional. They were using website trackers that collect user data on the site, which also tracks what a user fills out on a form. This would allow Cerebral to measure how users interact with ads on their website. But, since these tracking tools were made by their media companies, it also allowed them to have access to this information. While it most likely was not their intention to expose sensitive information about their patients to companies like Meta, they probably shouldn’t have been tracking their own customers. This information leak was only possible because they were tracking user data in the first place. This has happened before, with the company BetterHelp, who may have to pay a $7.8 million fine. While it is good that they might be convicted, fines for massive companies should be much bigger. Instead of millions of dollars, fines should be a percentage of total revenue in order to keep companies from doing such a terrible thing.

OpenAI says that this new model will be even more creative and collaborative and have greater accuracy. OpenAI also says that the AI would now have less hallucinations, meaning the chatbot won’t derail as fast or as often. While this is a good thing, its execution is a little questionable. Microsoft's Bing chatbot has been reported to have been using GPT-4 from the beginning and we saw how that went. Bing chat has insulted people and has gone crazy many times to the people who have been using it. Even after all of that, Microsoft has fired their AI ethics team. I am looking forward to seeing how Bing would handle this and reporting it later.

Apple appears to be working on some kind of natural language model update for Siri on Apple TV. This natural language is currently only being used to tell jokes and it is unknown what could come next. The recent increase of chatbot availability has made voice assistants, like Siri, look bad.