News 3 mins read

Worried About the Ethics of AI?

AI gets all the press these days, with fears of it causing job losses or creating maniac robots, but there is more to worry about out there than AI.

Tim Holyoake Tim Holyoake

Because artificial intelligence is a hot topic these days it gets all the press, with fears of it causing job losses or creating a maniac robot that takes over the world.

But there is more to worry about out there; the ethics of all software usage should be front and center in our thinking.

In April of this year the European Commission issued a press release outlining seven essentials for achieving trustworthy AI. They are all laudable principles, attempting to ensure that accountability, transparency, safety, privacy and so on underpin the work of AI practitioners.

For example, on societal and environmental well-being the commission states that AI systems should be used to enhance positive social change, sustainability and ecological responsibility.

No-one would seriously disagree with any of the principles put forward by the commission. And many people will want to work actively to ensure they are achieved. But there are problems with the focus of the commission’s proposals. 

These problems arise from tying these principles solely to the renewed fashion in software development circles for AI.

AI isn’t a single, well-defined approach to designing computer systems. The roots of AI can be traced back to the 1950s and the Turing Test. Fashionable as a computing paradigm in the 1990s, AI embraces many different approaches to problem solving. These include pattern-based methods like those found in machine-learning systems, as well as systems based on rules and logic.

The only commonality between different AI approaches is that they run on computing hardware not dissimilar to the one you’re viewing this article on, using algorithms written in a computer programming language.

It’s therefore the development, use and misuse of computational algorithms and the data they rely on to feed them that our ethical focus needs to be trained on, not on a single hot trend in computing.

If our ethical focus is on AI alone, we’re more likely to miss all of the other abuses perpetrated by computer algorithm misuse. For example, take Bitcoin mining – a decidedly non-AI computing technique. This is currently responsible for up to 22.9 million tons of carbon dioxide being pumped into the atmosphere every year. That’s almost equivalent to the carbon dioxide emissions produced by Sri Lanka.    

We therefore shouldn’t worry about the ethics of AI in isolation. If we do, we’ll simply pin all the shortcomings of the software industry on a currently fashionable approach to designing systems, neatly sweeping wider ethical issues under the carpet.

All software usage needs to be considered. Get that right, and the seven essentials outlined by the European Commission for achieving trustworthy AI will take care of themselves.   

For more information on analytics and artificial intelligence please click below.