OK Computer? Artificial Intelligence and TV Diversity

Being in China, I’m constantly discussing and covering news about Artificial Intelligence (AI) . And although “Alexa” doesn’t work so well in China because of the great firewall, I still love the fact that because of AI I can just shout and Alexa will pick a playlist that perfectly fits my mood. Somehow, I feel like my standard of life is improved. A few months ago, however, In October 2018 a small story broke about Amazon’s use of AI technology in hiring staff.

Don’t feel bad if you didn’t hear about it, most people didn’t.

But the story potentially has huge consequences and should send shivers down the spines of everyone who cares about diversity in television.

In 2014 a team at Amazon.com started developing a computer programme to review job applicants’ applications with the aim of mechanizing the search for top talent. The Artificial Intelligence (AI) employed in the computer programme would be devoid of human emotions and prejudices and objectively evaluate potential candidates.

In many ways you would think this is exactly what people who campaign for employment for people of more diverse backgrounds have been clamoring for. No more bosses just emplooying their friends and employing in their own image.

But by 2015 the bosses at Amazon started to realize that something was going wrong.

The AI programme was weeding out all the women.

The computer programmers had given the AI system over ten years’ worth of successful employees' CV’s and so the AI programme had “learnt” that one of the biggest indicators of whether a candidate would be successful in the future was whether the candidate was male. Hence all women were got rid of straight away. Far from getting rid of prejudice the computer programme only amplified the prejudice of previous employment practices combined with the possible subconscious innate prejudices of the programmers (who happened to be predominantly men).

For people interested in diversity in television this case potentially has far wider consequences than just employment practices at the online retailer.

In May 2018 Broadcast magazine published a piece explaining how AI could be used in the television industry to analyse scripts and decide which ones get green-lit or not, or at the very least which ones commissioners get to see from the mass of scripts which are submitted.  

AI could also be used to tweak scripts changing the gender or race for example of a particular character if its machine learning has taught it that a woman or white man are viewed more favourably by an audience.

In 2017 the BBC announced it was partnering with eight British universities to use AI to "better understand what audiences want from the BBC."

The publication Engadget reported that “The BBC hopes machine learning can help it build ‘a more personal BBC’ with tools that could allow employees to make informed editorial and commissioning decisions.”

Looking at the example of Amazon the potential pitfalls of using AI to decide who to employ and what programmes to commission are obvious and scary. But more importantly the problems might be far harder to spot than the Amazon example.

In America right now people are so worried about the potential prejudices being imbedded into AI models that journalists and lawyers are arguing that AI programmes should be open to be reviewed and challenged for prejudice. That means people should be able to independently analyse the computer codes that are used to teach the AI programme and make sure there aren’t any assumptions baked into the programme that might discriminate against certain groups. Just as any human decisions can be challenged in court if someone is suspected of sexism or racism for example computer programme should also be subject to the same type of challenges if prejudice is suspected.

I love AI – it improves my life in so many ways and I am not arguing for a return to a world where it doesn’t exist. Nor am I arguing for a return to a world where scripts are green lit and people are employed just on a nod and a wink. However, I personally side with the US lawyers and journalists who are making sure that we subject all decisions to scrutiny. Whether those decisions areat is made by a humans or ca computers.

Now after writing this blog post I need to relax. - Alexa play me some 90’s R&B