October 3, 2017

Humans and AI: the magic combination

Charles is a former PPA Business Journalist of the Year, specialising in AI and Fintech. He began his career as editor of Euro Business Magazine and now writes on the rise of AI and algorithms in business.

Smart Platform blog
Who is better at guessing a hit song, a machine or a human? German artificial intelligence expert Professor Matthias Seifert of IE Business School thought it was a good test of AI, and invited 180 people, half experts, half amateurs, to predict the Top 100 positions of singles by old and new artists. They competed against modelling software.

The results were startling. He explained:

“Making a prediction about a song by an established artist is a ‘well-structured’ problem... Here, we found that in a straight contest of man versus machine, the machine tended to win.”

But, he added: “the best results came from mixing human and computer predictions.”

For unknown artists, humans were the best. The machines were starved of data and struggled to compete. But, once, more: “combining computer and human predictions produced the best results.”

There are good reasons for this. Machines can crunch numbers in a way no humans can match. Their statistical modelling is profound. But humans can deal in “non-linearities” - situations where data is missing, unstructured, or falls outside normal patterns.

Combine the two and you get the best of both. 

A White House report on AI came up with the same verdict. Published in 2016 at the behest of Barack Obama, the report consulted a veritable Who’s Who of tech experts across the United States. Again, the researchers asked how artificial intelligence would work with humans.

It concluded: 

“In many applications, a human-machine team can be more effective than either one alone, using the strengths of one to compensate for the weaknesses of the other. One example is in chess playing, where a weaker computer can often beat a stronger computer player, if the weaker computer is given a human teammate—this is true even though top computers are much stronger players than any human.” 

The report cited radiology a great example. In a test, an AI-based approach had a 7.5 percent error rate, where a human pathologist had a 3.5 percent error rate. Yet a combined approach, using both AI and human input, lowered the error rate to 0.5 percent. That’s a sensational verdict in favour of human-machine co-operation.

How industry is using AI

In practical commercial situations we are again and again seeing AI work in harmony with humans to produce outstanding results.

Swiss-based Gamaya has raised $3.2m in funding for its AI project. It has equipped drones with hyperspectral cameras that monitor water and fertiliser use, crop yields, and pests, and applies an AI algorithm. Farmers can decide what action to take. The result is lower burden of duties for humans, and higher yields.

In supermarkets, AI forecasting engines are being used to automate the ordering of goods. Morrisons is working with AI agency Blue Yonder to automate ordering of 26,000 products in all its 491 stores. This was back-breaking labour, now done by machine.

Humans are liberated to focus on the parameters. For example, how acceptable it is to run out of a product. Shoppers tend to get annoyed if baby formula is out of stock, but less worried by an absence of aubergines.

The results? Morrisons has reduced shelf gaps by 30 per cent and reduced stock holding in store by 2-3 days, yet with increased availability.

More jobs, not fewer

When humans work with AI the results are more jobs overall, although in different roles. Boring jobs disappear. Laborious work is automated. Humans are elevated to higher-impact roles.

Research by Tata Consultancy Services on global trends found AI is causing a surge of jobs in retail. Executives in the sector told Tata that investment in AI helped them achieve an average of 19 per cent increase revenue in 2015 to the year before. They also report an average cost reduction of 15 per cent related to AI initiatives.

Capgemini ran a global study of 1,000 executives at large corporations worldwide. It found three-quarters of UK organisations implementing AI have created new jobs as a result. It also highlighted the positive impact of AI on business growth, as 68 per cent of UK organisations reporting a 10 per cent uplift in sales, directly tied to AI implementation. 

A report by Accenture estimates AI can increase productivity by 40 per cent or more by 2035. That’s $4.7 trillion in value.

The old narrative positioned AI as a challenger to humans. It would put them out of a job. Or undermine their worth in the market. 

The truth is the reverse. AI is an enabler. It unlocks new fields. And the peak results come when humans and AI work together.

Blog

More blog articles