AI must develop for the greater good and cut government admin

Role of diversity

Speakers at the Westminster eForum also discussed the importance of diversity. Wendy Hall, professor of computer science at Southampton University, said “bias in, bias out” is equivalent to the adage of “garbage in, garbage out” in data management, where poor-quality data at the point of collection leads to poor data output.

For TechUK, diversity needs to span the whole of business, said Daley. “We need more diversity in the AI community and across management to apply AI within business,” she said.

Some businesses risk losing out because they are simply too far behind in digitising their internal processes, said Daley. “Lots of organisations are still using paper invoices and are not on their digital journey,” she added. This limits their ability to make the most of AI in their business.

Hall described AI today as the the fourth wave of AI, which has developed thanks to algorithms that can now make the most of available processing power to analyse vast pools of data. But she pointed out that current data legislation may limit the ability of startups to exploit this data.

“This wave of AI is all about how we deal with data,” she said. “China doesn’t have GDPR [the EU’s General Data Protection Regulation] and there are billions of people in China and they will be able to hone their AI algorithms better than we can. How do small companies get access to data to hone their algorithms?

“Our main recommendation is that there should be a legal and ethical framework to enable access to confidential government data, such as health records.”

TechUK’s Daley added: “Getting data right for AI needs data to flow from the UK to Europe and the rest of the world in a post-Brexit world.”

Chris Melhuish, director of the Bristol Robotics Laboratory, argued that there needs to be a culture change in academia, which has traditionally been focused solely on pure research, to encourage the creation of new businesses to exploit technologies.

Nesta’s Mulgan said that beyond establishing policies and funding to support an AI-empowered society, governments should also look at how AI can optimise their own inefficiencies. “Most of the investment of AI in the last 50 years has been from government, but there are no systematic programmes to use AI in government,” he said.

Southampton University’s Hall added: “Government needs to be a shining light in AI adoption. It is all about modernising the approach to IT, and government needs to be a model of best practice.”

The discussions and presentations at the eForum explored topics that experts have been investigating for a number of years. There was a sense during the seminar that AI will augment jobs, rather than replace them. But AI could completely remove certain types of job, which means people will need to acquire new skills.

In 2016, PwC’s McCargow blogged: “Developing the right skills to ensure we can continue to innovate is important. One school of thought is to equip the workforce of the future purely with digital skills, but because AI has the potential to democratise access to technology and code for us, humans will need to focus on creativity and critical thinking.”

This was an issue that Andrew Joint, managing partner and commercial technology partner at law firm Kemp Little, covered in his presentation, warning of the risk of automating too many tasks.

Joint described how, when he started out as a young lawyer, his job involved reading masses of case histories – a task that can now be automated using an AI reading algorithm. But lawyers gain a greater understanding of case law by reading this material, he said, and there would be less opportunity to gain this knowledge if the process was fully automated.