<img height="1" width="1" style="display:none;" alt="" src="https://px.ads.linkedin.com/collect/?pid=2323978&amp;fmt=gif">
Skip to content

Not Just Black and White – AI Positive at ‘The Zebra Project’

"When you hear hoofbeats, think of horses, not zebras". This was the advice originally given to medical interns in the 1940s, encouraging them to normally trust the first answer to a problem that came to mind when diagnosing a patient. In the 21st century, you’ll most often hear it doled out by advocates of the ‘path of least resistance’, discouraging anyone from pushing the envelope when a simpler or easier way of thinking and doing presents itself.

But perhaps now is the time to start thinking ‘zebras’. The UK’s productivity still lags well behind our European neighbours; tired ways of thinking in industry have been having a detrimental effect on results for decades. So, when it comes to realising the benefits of automation and robotics, we should be looking for innovative new ways to usher in positive change rather than treading the familiar sceptical path.

Taking the road less travelled

That’s why it was so enjoyable to speak on the panel at the latest instalment of Taylor Vinters’ ‘Zebra Project’. This series of expert peer-to-peer discussion and workshops covers all things ‘future of business’, and this particular afternoon was set aside to ‘keeping humans at the helm’ of automation and AI developments.

It’s an important point; much of the worry and stress that comes from debate on robotics/AI is based on a fear of mass job losses and the ‘replacement’ of humans in the workforce.

Thankfully, as the keynote speakers and colleague discussion showed, it’s just not as simple as that, and there is far more room for optimism than pessimism when it comes to working with AI. Here are the two ideas that I’m happiest to steal from the brains on show:

Moravec’s paradox is a beauty

We’ve known for years that robots can use their enormous computational power and machine learning capabilities to do things that humans simply can’t. A much-publicised example was the rapid success of Google’s AlphaGo and Alpha Zero Go/Chess machine-learning software against leading humans and human-built software.

We thought it would take a lot longer for AI to master these supposedly ‘difficult’ tasks, but the timeframe was a fair bit shorter than that! Now we get to enjoy the real-world implications of this ability, such as IBM’s Watson in the healthcare space helping to identify and treat cancer faster and more effectively than human doctors.

On the other side of the coin, tasks scientists first thought would be simple, are still baffling robots and AI! It takes a vast amount of resource to do things that we humans find incredibly easy such as turning the page of a book or putting a plate on a shelf, not to mention more philosophical points, like explaining the difference between a chair and a table, or why one sunset is beautiful and another isn’t.

This is Moravec’s paradox; we really do know more than we can tell, which means humans still have an awful lot to teach AI, and have many areas where their performance will outstrip robots for years to come.

AI complements as well as competes

There are many ways that we can look to maximise the value and deployment of technologies without disadvantaging people. Many smaller businesses are using AI and RPA to automate time-consuming and dull processes; this frees up workers to focus on more value-adding tasks, but also reduces the error rate in the original ‘drudgery’.

In some automation use cases, we’ll see ‘recycled demand’, where automating parts of an industry leads to consumers re-investing their savings in the same industry – in the 19th century it was garment making, perhaps it will be legal clerking in the 21st century?

Ben Dellot, Associate Director of Economics, Enterprise and Manufacturing at the RSA, was the keynote speaker, and he provided a strongly evidence-based picture of the automation landscape as it lies now. A lot of the time, rampant cheerleading or doom-mongering, seizing on whatever stats the media is interested in that day, fails to give us a realistic understanding of what’s going on and an accurate timeframe of development.

It turns out, and I’ve been shouting about this myself for a few years, that there are very few ‘active adopters’ of automation and robotics. Many businesses still see automation initiatives as too expensive or confusing; some are still not even aware of what AI is. From experience speaking to c-level executives at various businesses, they ask questions like ‘so what should I do with AI?’ It’s not one thing! It’s a million different capabilities which will be paired with different technologies to achieve business aims.

This complementary approach is far more common than a straight swap of robot for human – and serves as a reminder that balancing and governing a human/digital workforce is one of the real automation issues decision-makers should be concerning themselves with.

All credit to the organisers of the Zebra Project; I’m looking forward to what the rest of the series has to cover!