Amazon AI recruiting tool removed for favouring men for technical jobs

Since 2014, Amazons specialists have been building computer programmes to review job applicant’s CV’s. The aim of the programme was to search and select the top talent, giving candidates a rating from 1 – 5 on how suitable they were for the role. The tool used artificial intelligence to go through all applications and select the top 5 which it thought was a good fit for the role and amazon would proceed to hire those top 5 applicants.

Not long after the machine learning specialists had created the programme, did amazon realise that the system were not rating candidates for their technical roles in a gender-neutral way. In 2015, Amazon had realised that their computer models were picking applicants by observing the patterns in previous resumes submitted to the company over the past 10 years. This was where they found male dominance across their tech department, which reflects the tech industry stereotype.

Amazons system had taught itself that male candidates were preferable for the roles and had disregarded resumes with the word ‘women’ in. Even though Amazon edited the programmes to make them gender neutral, there is no guarantee that the machines would not find other ways of sorting candidates that could discriminate.

After Amazons headcount had tripled, they decided to set up an engineering hub in Edinburgh. Here is where they developed the programme that used artificial intelligence to search the web and find candidates worth recruiting. 500 computer models were created to focus on specific job functions and locations. They were taught to recognise 50,000 terms that were found on past candidates resumes and the algorithms learned skills that there common across tech applicants. However, the machines were favouring applicants that described themselves using masculine language along with recommending unqualified candidates for roles due to there being problems with models’ judgement. With the technology constantly making mistakes and giving Amazon the incorrect data, they shut down the project immediately.

Along with Amazon, other companies are going ahead with using AI to hire and select well suited candidates. Even though this can speed up the recruitment process and save employers time, it can also be very risky. Businesses need to be careful about the data they feed the software as they need to ensure it is ethical and doesn’t cause them issues that Amazon have experienced. Ensuring the machines do not discriminate during the selection of resumes is also essential, otherwise companies can be seen to have a diversity problem. The tech industry’s stereotype will continue, and women will not be given an opportunity to succeed.