Recognition software, virtual assistants, and other AI based algorythms have a bias challenge to overcome. Products by Google, Amazon and even firms with profile in algorythms used for crime prediction have been found to arrive to biased conclusions for races, genders as well as places of abode.
The reasons algorithms behind AI become biased relate mostly to the way they teach themselves, as the data comes from a very broad range of constantly changing sources. The biased software is therefore very difficult to reverse-engineer, hence both Google and Amazon have abandoned some very advanced projects in this area.
Salesforce, one of the global leaders in cloud computing, may have a solution. Adam Caplan, Salesforce’s head of blockchain, suggests to structure and track the disorganized incoming data using blockchain:
“Blockchain can be really powerful in terms of how these algorithms are made and what’s influencing them, or if there is bias in AI to be able to understand how that came about.”