Microsoft to phase out public access to controversial facial recognition tool that identifies emotion

Microsoft is reportedly phasing out public access to a few of its AI-powered facial analysis tool, including one that claims to identify emotion based on videos and pictures of subjects.

Experts have long criticised such “emotion recognition” tools because there are no universal facial expressions and even if there are, they differ across populations and it is grossly unscientific to “equate external displays of emotion with internal feelings”.

Lisa Feldman Barrett, a professor of psychology at Northeastern University, told The Verge that AI-powered emotion recognition tools can detect scowls, “but that’s not the same thing as detecting anger”. Barrett has conducted a review into the subject of AI-powered emotion recognition.

Microsoft’s decision to phase out this particular tool is a part of a “larger overhaul” of the company’s AI ethics policies.

The company has updated its Responsible AI Standards, which were first outlines in 2019, to emphasise on accountability and “find out who uses its services and greater human oversight into where these tools are applied”.

So, what does this mean?

For starters, Microsoft will limit access to come features of its facial recognition services on ‘Azure Face’ and remove some others entirely. Users will now have to apply to use Azure Face and tell Microsoft exactly how and where they will be deploying these systems. Some use cases though, like automatically blurring faces in images and videos for example, will remain open-access.

Microsoft is also retiring Azure Face’s ability to identify attributes like gender, smile, age, hair, facial hair, and makeup.

“Experts inside and outside the company have highlighted the lack of scientific consensus on the definition of ‘emotions,’ the challenges in how inferences generalize across use cases, regions, and demographics, and the heightened privacy concerns around this type of capability,” wrote Microsoft’s chief responsible AI officer, Natasha Crampton, in a blog.

The company has announced that these features will no longer be offered to new customers from June 21 while existing customers will have their access revoked from June 30, 2023.

Now, here’s the twist. While Microsoft is retiring “public” access to these tools, it is going to keep using them in “at least one of its own products”. This product is an app named Seeing AI that “uses machine vision to describe the world” for visually-impaired people.

Microsoft’s principal group product manager for Azure AI, Sarah Bird, mentioned in a blog that tools that recognise emotion can be valuable when they are used “for a set of controlled accessibility scenarios”. And, it is not clear if the company plans to use these tools in any other product.

The company is also going to introduce similar restrictions to its Custom Neural Voice feature which customers can use to create AI voices from recordings of real people, also called audio deepfake. Bird said that while this tool has exciting potential in education, entertainment, and accessibility, it can also be used inappropriately to deceive people. Microsoft has said that it is going to limit access to the Custom Neural Voice feature to “managed customers and partners” and also ensure the active participation of the speaker involved when a synthetic voice is created.

Also Read: Meta, Microsoft, other tech giants form metaverse standards body, except Apple

Also Read: Facial recognition system planned at Delhi, Pune, other airports by Mar 2023: Govt