We know remarkably little about how AI language models work

Learn the total story right here. 

Deeper Studying

Google DeepMind has launched a watermarking software for AI-generated pictures

Google DeepMind has launched a brand new watermarking software that labels whether or not pictures have been generated with AI. The software, known as SynthID, will initially be out there solely to customers of Google’s AI picture generator Imagen.  Customers will be capable of generate pictures after which select whether or not so as to add a watermark or not. The hope is that it may assist individuals inform when AI-generated content material is being handed off as actual, or shield copyright. 

Child steps: Google DeepMind is now the primary Large Tech firm to publicly launch such a software, following a voluntary pledge with the White Home to develop accountable AI. Watermarking—a way the place you cover a sign in a chunk of textual content or a picture to establish it as AI-generated—has develop into one of the widespread concepts proposed to curb such harms. It’s a great begin, however watermarks alone received’t create extra belief on-line. Learn extra from me right here.

Bits and Bytes

Chinese language ChatGPT options simply received authorised for most people
Baidu, considered one of China’s main artificial-intelligence corporations, has introduced it is going to open up entry to its ChatGPT-like massive language mannequin, Ernie Bot, to most people. Our reporter Zeyi Yang appears to be like at what this implies for Chinese language web customers. (MIT Expertise Evaluation)

Mind implants helped create a digital avatar of a stroke survivor’s face
Unbelievable information. Two papers in Nature present main developments within the effort to translate mind exercise into speech. Researchers managed to assist ladies who had misplaced their potential to talk talk once more with the assistance of a mind implant, AI algorithms and digital avatars. (MIT Expertise Evaluation)

Contained in the AI porn market the place every little thing and everyone seems to be on the market 
This was a wonderful investigation how the generative AI increase has created a seedy market for deepfake porn. Fully predictable and irritating how little now we have carried out to stop real-life harms like nonconsensual deepfake pornogrpahy. (404 Media) 

A military of abroad employees in “digital sweatshops” energy the AI increase
Tens of millions of individuals working within the Philippines work as information annotators for information firm Scale AI. However as this investigation into the questionable labor situations exhibits, many employees are incomes under the minimal wage and have had funds delayed, decreased or canceled.
(The Washington Submit) 

The tropical Island with the recent area title
Lol. The AI increase has meant Anguilla has hit the jackpot with its .ai area title. The nation is anticipated to make thousands and thousands this yr from corporations wanting the buzzy area title. (Bloomberg)

P.S. We’re hiring! 
MIT Expertise Evaluation is on the lookout for an formidable AI reporter to affix our crew with an emphasis on the intersection of {hardware} and AI. This place is predicated in Cambridge, Massachusetts. Sounds such as you, or somebody you realize? Learn extra right here. 

Leave a Comment