The starting point for many conversations around AI is based on the fear felt by many over the future of jobs. There is substance to this, in my view, but there are certainly only a fraction of jobs, if any, that will not be affected by AI in the relatively short term. This has happened before. Indeed, it has happened many times. New waves of technology bring disruption to established roles as they did during the agrarian, and then industrial, revolutions (think Luddite reactions). There is no reason to believe that we have become more fragile than we were in the past and that society will not adapt.
Part of the fear comes from the consistent portrayal of AI in popular culture, culminating in sentient robots running around, trying to overthrow human masters in displays of raging anger and violence. This is not going to happen anytime soon. There are too few positive films about AI robots just getting on with making life easier, as my colleague Prof John McIntyre has noted, so too many of our mental pictures are more negative than we are ever likely to experience. For me, there are legitimate dangers in the future development of AI (where we all end up like the do-nothings floating round in Wall-E), but ending all jobs while we still want to work is not one we should be concerned with.
AI is just another example of how machines can help us humans do more, in ways we like. Cars and bikes help us go further, faster than walking alone. Some human roles were disrupted when we started using those machines, (farriers became a lot less common place) but overall, the experience was good and, as nothing like this happens overnight, people were able to retrain or change profession over time. As a society, we will need to help those similarly disadvantaged but AI will be a net benefit, that is clear.
The way most of us will experience, in fact already are experiencing, AI is through mundane areas of life that we will be glad to have disposed of.
The term AI is actually a pretty broad one that covers not just the robots-as-gods nightmares but also Machine Learning, Natural Language Processing, Recommendation Engines (like Deezer and Spotify), Process Automation and a host of other areas. AI is an elastic band term that has been stretched to cover a lot of different technologies. It’s in these broader areas that we are already seeing AI made real. It’s perhaps easiest to put the immediate uses of AI into two categories: automation of processes that are done repeatedly and enhanced decision making based on analysis of big chunks of data that humans would have a hard time crunching through.
It’s perhaps easiest to put the immediate uses of AI into two categories: automation of processes that are done repeatedly and enhanced decision making
Most of us have probably used a chatbot. If you have ever logged on to the website of a big company and engaged in chat to ask questions, book a delivery or get something remedied, there is a reasonable chance that at least some of your conversation was with an AI-powered bot rather than a person. The automated bots allow more enquiries to be answered more quickly, and for more hours of the day than humans can manage. It’s cheaper for the companies to operate and there is a better service for us, too.
As with many big companies, local governments have started to adopt similar products. Aylesbury District Council is using chatbots, as one pioneering example. This is a good case of how technologies in use in other industries can readily be applied to government and the public sector without the risks associated with newly developed technical solutions. Sentisum is a UK-based start-up (backed by Y-Combinator, among others) looking to use its AI-based products to improve customer experiences on line and has government departments in its sights following early success in the insurance and retails sectors.
One friend of mine (they know who they are!) has developed an odd obsession with AI tools in waste collection, another key local government service. AI-based robotic systems (like the ones developed by Clarke that has been tested by a council in the eastern region of the UK) are being used to help sort waste from recycling bins. We know that when we tell people they don’t have to sort their recycling, they will put more in the bins. So, making that possible by having automated sorting is a good way of increasing recycling rates reusing materials that we should all welcome. Such waste management schemes could be even more effective if bins had sensors to say when they needed collecting, so AI systems could best plan routes, cutting costs and urban pollution; again, another benefit for everyone in society. Such route planners and decision making programmes are already at work in delivery and logistics companies – they can just be repurposed for local government and public service needs.
Other projects I have seen are looking at: using social services data to help predict domestic abuse; sentiment analysis of Twitter to help prioritise council repair work; models for social care using demand planning; maximising the amount of new housing possible in an urban environment; and data-driven educational needs planning tools based on massive social data sets. These are all examples of current AI technologies being used to solve government and public-sector needs.
Just because some of these tasks might be mundane-sounding, does not mean they will not be big in their effects (Deloitte estimates they will save the UK government £17bn pa by 2030) and that means there should be opportunities for entrepreneurs in B2G (that’s business-to-government) and investors. In these ways, too, AI in government will be just like AI everywhere else.
Full article available on Linked In