AI, the future intelligence for learning or more biased than a human?

Copy of Diversity with inclusion- linkedin (1)

by Johanna Beresford, CEO

Each day there’s a constant stream for data that feeds and builds machine-based learning. Should we consider this scary? Powerful? or some varying combination of the two? The moral code of AI has been discussed extensively over recent years, yet we have not stopped building these ever-intelligent machines. Some experts say, we are beyond the point of return and AI and robotics is here to stay – there’s no ‘off’ button.

As I wake up this morning, reading the news, browsing and engaging across several different social media platforms for both personal and professional reasons, I’m constantly reminded of the intelligence build- whether that be LinkedIn, BBC News, Instagram – every single item I’m fed, are all pieces of content I’m interested in. As such, I become more and more drawn into the next article and want to find out and learn more. This hasn’t happened by accident, the things I see, are specifically selected to prick my interest and I enjoy (in the most part) this content that’s recommended to me, I’m inquisitive.

These machines to date, have been developed to recommend learning that we, are highly likely to be interested in and engage with at an individual level. Learning pathways built into LXPs are based on your previous behaviour, as well as things you indicated an interest in and the intelligence around your demographics -in my case –  as women in her late 30s with 3 children, I want to consumer content around work life balance, flexible working, potty training etc. But what if these aren’t my areas of focus? What if I’m not a ‘typical’ woman (“a ‘typical’ woman” – now that’s a topic for another discussion…). We are all different, that’s the beauty in the human race. We know and accept that as human’s, we come with our own individual biased, but we must also be aware that these biases can then become the foundation of machine-learning and then that bias is rapidly perpetuated.

Mo Gawdat, Ex Chief Business Officer of Google X and Author of “Solve for Happy:  Engineering your path to Joy’ recently said “We are at a junction of our history where more technology may not be what we need” and believes by the year 2049, machines will be one billion times smarter than us. He likens this to comparing a fly to Albert Einstein and then sets the question – “will there still be a need for the fly?”.  Whilst many of us will have grown up to see several single technological developments over the last 30 years such as the development of personal computers, the internet and smart phones, we are currently at the centre of the development of multiple technologies at one time – something mankind have never previously experienced. Mo further expresses that in regards to robotics and AI, we must remember, this is not about simply building a new machine but rather, creating a new form of being that is creating its own intelligence and is currently in its infancy. Pretty similar to a toddler – constantly learning and trying to figure things out, until it understands and then moves on. Mo also theorises that AI will grow emotions and values through its pattern recognition from the data gained from us, therefore we have to be mindful of the content and language we use – whether that be on social media, in emails, internet search and even WhatsApp messages. If machines are learning from how we are today – what emotions and values do we regularly express? Are we generally optimistic, kind-hearted, empathetic and considerate? Or is it easier to find examples aggression, violence, dismissiveness and negativity?

Is there a different way of building AI? One that encourages us to be and celebrate our difference, to interact with learning that isn’t typical our demographic profile. Can we create an AI that recognises bias, acknowledges it and suggests alternatives? Through thinking about the value of difference, rather than aggregating commonality – could we build a more sophisticated, moral AI to develop learning?

Related Posts

Tags

Recommended