A shocking new report reveals how tech companies are acknowledging the fact that their algorithms can perpetuate serious concerns such as discrimination.
One common example is an artist named Stephanie Dinkins who has been an industry leader for many years. She is known not only for her stunning creations, but also for her curated selection of exceptional products that combine her art with contemporary technology.
She has won numerous awards and has made her name proud with all sorts of groundbreaking innovations, and has won prize money as well.
Over the past decade or so, we’ve seen her make the most of AI technology to really show Black women in every light. From happiness and celebration to sadness, pain, grief and worry, you name it and she does it.
Did I mention the long list of interviews she’s had, where she’s been questioned about her work, her talents, and sheer masterpieces. But wait, not everything is positive.
Critics were seen saying that her love for AI that portrays realistic emotions could be as alarming as it is fake. For example, her proprietary algorithm finds that humans are rendered in shades of pink, plus they’re covered in black cloaks.
She expected one day to be called and questioned about her art. Furthermore, she says, she was convinced that her first work was just an experiment and that it would get better and better with time.
According to her, it’s not difficult to get the AI tool to generate the desired image through a few specific text prompts. True to her word, the machine gave her everything she needed, but the distortion was always there.
For example, when she uses words like African American and Black woman interchangeably, the characteristics are very different. Everything gets a little skewed and problems occur more often than expected.
For her, it’s all a bit of a shock, but she’s gotten used to it, but she’s hoping that change will happen soon as discrimination seems to be happening at a higher rate than usual. For her, system-generated biases are a worrisome and worrying factor.
She hopes that one day the system will figure out what exactly black women are so that she doesn’t harbor nuanced feelings.
But you’d be surprised that the famous artist isn’t alone in raising tough questions about the twisted relationship between race and the world of AI. There are many people in this world who have admitted that AI intelligence is racist and biased.
The problem is that the data provided to train the algorithms in creating images and the machines on which such programs work belong in this category.
AI technology may be ignoring or altering the actual text prompts presented in terms of how black people should be portrayed in the photo. On the other hand, in other cases they may be censoring both cultures and their pasts.
We have seen this topic become a hot topic for years. From technology that recognizes faces to the difficulty of understanding someone else’s unique speech patterns, the list can go on for days. And a growing number of surveys and studies are raising questions related to both prejudice and equality.
Now the question is what is the solution. Top tech giants are working behind some of the biggest AI generators. From Midjourney to OpenAI, they all admit they have problems. And they all vow to make the tools at hand better. They say stigma is a big problem plaguing the entire industry and change is needed to avoid it.
OpenAI was seen in a recently published interview saying it is trying to reduce the prevalence of such biases while improving performance at the same time. But it’s not entirely clear what exactly the company does and how many members of the team have been delegated to undertake such a huge and comprehensive task.
For now, this is definitely not something that can be ignored or fixed overnight, so we’ll have to wait and see. What do you think?
Image: DIW. Story H/T: The New York Times
READ NEXT: New study reveals how people feel about AI making moral judgments
