Search This Blog

Tuesday, December 3, 2019

gender bias in chatbots too !


Robots are quite interesting ! (this post is more on bots!) ~ in Rajini starrer Enthiran (Robot) – Chitti, the humanoid robot would read a full book by just scanning it in his face for a couple of seconds.  That perhaps was a thought that Sujatha wrote in ‘Mr Munsamy oru  1.2.1’ – in which a roadside rickshaw puller acquires super memory by an injection that were to lost only for a short period !! I felt ‘Enthiran – the robot’ was a poor cousin of his classic ‘En Iniya Iyanthira and Meendum Jeano’ - the little Jeeno, the cute  robotic dog in the story  woven subtly around a dictator who keeps the Nation under tight grip.  The real identity of the Head when known at the end  was the real classical revelation.  The pet robot dog which can think beyond humans, assists the dumb Nila in search of her spouse Sibi into bigger things.  Towards the end, the cute exceptionally dog loses it memory and back up and fades away !I felt very sad reading of its end  -  illogical it might sound, it was after all a story – yet  !!

Downunder, a bot, Christened Mia, or My Interactive Agent, the AI-backed virtual assistant  made waves that it can respond to 300 of the most common customer inquiries made during the home loan application process.  As you make online transactions on web, a drop down figurine might introduce him/herself and ask  whether any assistance is required .. … in all probability it is no human interface !

A chatbot is a piece of software that conducts a conversation via auditory or textual methods.  Such programs are often designed to convincingly simulate how a human would behave as a conversational partner, although as of 2019, they are far short of being able to pass the Turing test.  Chatbots are typically used in dialog systems for various practical purposes including customer service or information acquisition. Some chatbots use sophisticated natural language processing systems, but many simpler ones scan for keywords within the input, then pull a reply with the most matching keywords, or the most similar wording pattern, from a database.  The term "ChatterBot" was originally coined by Michael Mauldin (creator of the first Verbot, Julia) in 1994 to describe these conversational programs. Today, most chatbots are accessed via virtual assistants such as Google Assistant and Amazon Alexa, via messaging apps such as Facebook Messenger or WeChat, or via individual organizations' apps and websites.

Beyond chatbots, there are  Conversational AI used in messaging apps, speech-based assistants and chatbots to automate communication and create personalized customer experiences at scale. Vodafone, like any other storied telco giant today, has been urgently seeking ways to trim costs. Not only are its revenues under pressure from smaller upstarts, it must also spend heavily on upgrading its network infrastructure to handle faster, 5G data speeds.

One cost-cutting idea that Vodafone has chased is to automate certain internal tasks carried out by staff with simple bots and, in some cases, more sophisticated chatbots. The painful result is fewer people on payroll. Vodafone has, for instance, deployed so-called “robotic process automation” bots to automate back-office tasks like data entry. As a result, Vodafone said that in the first half of 2018 it laid off 900 people. .. .. for sure, Vodafone is not alone ! ~ we have heard this in Insurance circles too .. that some backend processing is planned to happen using bots.

Here is an interesting piece read in Forbes .. .. “ Meet Amy. And also Debbie, Inga, Mia, Erica, Eva and Cora”. These aren’t the members of a new, all-female rock group, but names that several large banks have been giving to their automated digital assistants. So-called chatbots have become a useful cost-cutting tools for companies with large subscriber bases (think banks, insurance firms and mobile phone operators). As they replace human call-center workers, such bots will help save banks an estimated $7.3 billion in operational costs by 2023, Juniper Research predicts. But the proliferation of bots with female names raises questions about whether they might also perpetuate gender stereotypes, particularly around the notion of women in the role of assistants. That criticism has already been levelled at Amazon’s digital assistant Alexa and Apple’s Siri.

A Forbes analysis of Europe’s 10 biggest banks ranked by assets shows that at least three have deployed chatbots with female names on their websites and apps. One  chatbot Marie, available to retail customers on Facebook Messenger, was given the name "because it conjures up an image of someone who is helpful and friendly," Female chatbots abound in other regions and industries. Mia, the chatbot released by Australian digital bank UBank, was described by the company earlier this month as “empathetic,” “fun” and “a little bit cheeky.”  

IPSoft, a New York-based software company that sells chatbot technology to banks like Sweden’s SEB as well as mobile network giant Vodafone, has its own white-label version of a customer-facing chatbot, named Amelia. IPSoft’s CEO Chetan Dube denied that the chatbot’s name perpetuated stereotypes, when asked by Forbes during an interview in December, and said it instead highlights “the thought leadership that is represented in females."  "She was the first female aviator that tried to go around the world,” Dube added, referring to the 1930's aviator Amelia Earhart.  Forbes revealed earlier this month that Vodafone was measuring the success of its chatbots on how many staff could be replaced by the software. While that may be an uncomfortable metric, the more worrying consequence of chatbots, according to four industry experts questioned by Forbes, is the risk that they could reinforce certain stereotypes.

“Gender bias is an increasingly serious issue in chatbot design, especially for voice-based chatbots,” says John Taylor, CEO of action.ai, a British startup that makes chatbot software for banks and travel companies. “These assistants often perform tasks that many view as menial.”   Vitor Shereiber, a language specialist at the German language-learning app Babbel, says that focus-group testing might lead companies to assign a gender to a chatbot on the notion that it makes customer feel more comfortable. But, he adds, bots could spread unrealistic expectations of how women should present themselves professionally, just as photoshopped pictures have done for women’s perceptions of their bodies. Part of the challenge for companies is finding a balance between automating customer service without putting customers off. PwC recently described chatbots as being able to “massively enhance customer delight and loyalty” because of their “personal touch.” 

Interesting !

With regards – S. Sampathkumar
3rd Dec 2019.

No comments:

Post a Comment