Connected: Do algorithms dream of electric sheep?


A look into the implications of machine learning, chatbots and algorithms in our working lives.

Computers have been a part of our working lives for a good thirty years now, but we’re on the cusp of another massive leap forward in technology. Many work tasks have become automated, but humans have remained an essential part of completing any process. With AI and machine learning, many of these interactions will be able to be outsourced to computer programmes, with less need for human involvement.

Rise of the chatbots

A chatbot is a programme designed to answer simple questions. They’re becoming increasingly popular in customer service. They pick up on keywords in messages from human users and are able to provide answers based on what they determine you’re saying. The algorithms are becoming more and more sophisticated as programmers feed them more data.

It sounds great in theory – take out some of the mundane and repetitive tasks from our work days and let computers do what computers do best. Efficiency all around.

Ask your new HR chatbot how much leave you are owed or other basic questions around pay and entitlement. While these kinds of functions are relatively straightforward and minimal, the potential scope of AI is much broader. Are you comfortable disclosing harassment to a bot? What does it mean for employee privacy?

Look Mum, no hands

While bots are often made of strict rule sets and systems, AI are prone to flying wildly past predicted behaviours. Facebook recently shut down an experiment when two AIs invented their own language to communicate rather than using plain English. While an interesting result, this meant the chatbots weren’t going to be useful for their intended purpose – buying and selling products to humans in the Facebook marketplace.

There are also ethical issues with machine learning, including how they’re programmed. Machine learning or deep learning relies on large datasets to determine patterns, and how this data is harvested and used can have implications on privacy. Do you want your habits and movements to be used to predict your future behaviour? It starts to sound very creepy, even if there isn’t a human being directly viewing this information.
Humans have bias, and when we design we need to be aware of what bias we are encoding into our creations. Algorithms can have large blind spots, just like their human counterparts. We need to programme for empathy as much as for efficiency, or we are replicating systems that hurt people as we automate.

Futureproofing the future of work

The proliferation of this technology and the speed at which it’s entering our lives highlights the need for both workers and unions to think about how this will affect us, both on an individual level and across industries. The dissolution of entire roles is always something that must be approached with care, and with the rise of AI this is a very real possibility. Careers can disappear overnight.

We can’t halt the advance of new technology, but we are in a position to influence its design, development and implementation in our work lives.