Amazon.com Inc. (AMZN) is one of the largest investors in artificial intelligence capabilities. Despite their immense benefits, AI tools are not perfect and sometimes go wrong. Here are a few instances when Amazon’s AI tools goofed up and made a mess for the company.
Misogynistic hiring tool
On Wednesday, Reuters reported that Amazon had shut down an internal hiring project after its AI systems gave preference to men over women for software and technical jobs. The systems were simply following historical data patterns on resumes, which indicated a large percentage of male employees in the technology sector, and ended up downgrading women over flawed assumptions.
Although Amazon tried to make corrections to the tool, it was not sure that the system would remain free from all kinds of biases. The tool was used only in a developmental phase and not independently and it was never implemented on a larger scale. The project was unable to identify strong candidates and due to this as well as other reasons, Amazon decided to abandon it completely.
A spokesperson from Amazon issued the following statement, “This tool was never used by Amazon recruiters to evaluate candidates.”
Facial recognition gone wrong
Amazon’s facial recognition tool Rekognition misidentified 28 members of the Congress as illegal suspects in a test conducted by the American Civil Liberties Union (ACLU), a group that had opposed the project from the start.
Although Amazon claimed that settings used in the test were incorrect and that the tool can aid in crime prevention, the ACLU and other civil rights advocacy groups continued to rally against the project over concerns of unfair surveillance and retribution. The stiff opposition led the Orlando Police Department to pause its use of the tool for the time being a few months ago.
What were you thinking, Alexa?
From accidentally recording private conversations and sending them to contacts to unexpected eerie laughter, Amazon’s Echo device and its virtual assistant Alexa have many gaffes to their credit – some funny and some plain creepy.
Not alone
Amazon is not alone in the world of AI mishaps. Two years ago, Microsoft (MSFT) had to clean up a steaming mess when its chatbot Tay went off the rails and began posting inflammatory comments just a day after its launch. Facebook (FB) reportedly had to shut down one of its AI programs after two bots named Bob and Alice went rogue and created their own language.
So, what’s next?
It’s no secret that Amazon continues to invest millions of dollars in the development and adoption of artificial intelligence capabilities. From robots in warehouses to delivery drones, the online retailer uses AI in ample proportions to run its operations. These investments and their returns have helped push Amazon towards the trillion-dollar mark and when weighing the wins against the losses, these incidents seem like tiny misses.
Amazon and its peers in the tech industry remain popular in terms of stock choices due to their increased investments in artificial intelligence and machine-learning which are seen as good drivers of growth in the coming future.
Face-Off: Amazon’s facial recognition tool really messed up this time