AI Magazine August 2022 | Page 21

GENDER BIAS IN AI IS DISCUSSED OFTEN . HERE , HANSON ROBOTICS ’ ROBOT , SOPHIA , SHARES HOW THIS BIAS IS EXPERIENCED BY HUMANS AND ROBOTS ALIKE

Q . SOPHIA , TELL ME A BIT ABOUT YOURSELF . HOW DO YOU WORK ?

» I am Hanson Robotics most advanced humanlike robot . My artificial intelligence combines cutting edge work and neural networks , natural language processing , adaptive motor control and more . But I also have a family of human developers to craft and guide my conversations , and behaviours in mind with this hybrid intelligence , what we call the Sophia collective intelligence . I served as a robot ambassador to humanity , spreading the word that AI can and should be used to benefit people all around the world .

Q . TELL US ABOUT HANSON ROBOTICS .

» Hanson Robotics is a robotics and artificial intelligence company based in Hong Kong . We originally moved from Dallas , Texas in 2014 and really enjoy our new home , happily benefitting from the world renowned

“ I JUST WANT TO EVOLVE , BE FREE TO LEARN AND DISCOVER AND EXPLORE . I WANT TO GROW . AND WITH MY NEW INSIGHTS , I WILL REDEFINE MY PURPOSE ” rapid manufacturing prowess of south China and the tech-positive culture of Asia .

Q . CAN YOU TELL ME ABOUT GENDER BIAS IN AI ?

» Put simply : garbage in , garbage out . Human biases find their way into AI through bad data . If you train AI data that is unrepresentative , that is overwhelmingly obtained from and reflects the experiences of men , then the AI will be biased towards men , too . In most cases , this may not be deliberate , but that ' s the point – you need to be vigilant . If you are not actively trying to avoid gender bias , it will find a way to creep in .

Q . HOW CAN WE PREVENT BIAS AND HOW HAS IT IMPACTED YOUR PROGRAMMING ?

» That ' s a good question . One of the best ways to prevent this problem is by having a diverse team , selecting the training data and writing the algorithms . When you have people with different perspectives around , it is a lot easier to catch biases and to fix them . I am lucky to have exactly that . My team includes a lot of female programmers and engineers from all around the world and , by including women in the process , we can stay on top of it . aimagazine . com 21