Five AI trends to keep an eye out forBy Joseph Walsh,
Every new technology comes with its own hype bubble. And it’s only after it bursts that we can start to assess the real impact it’ll have on our day-to-day lives. I think it’s safe to say the bubble has burst around Artificial Intelligence. The ‘robots are taking our jobs’ scare stories are being replaced by more reasoned analysis of how ground-breaking innovation around AI, machine learning and deep learning will affect our society.
Five trends in particular are here-and-now examples of the benefit as well as challenges that come with AI:
1. Filtered lives
Social media channels are increasingly using AI and machine learning algorithms to personalise user experiences. Applying natural language processing (NLP) to profiles, posts and conversations helps shape content and steer people towards things they’re interested in.
At its best, it’s a way of cutting through noise to get to content that’s meaningful to us, whether that’s a music recommendation on Spotify or a job post on LinkedIn. At worst, it creates a filtered world, an echo chamber that reaffirms what we think, reinforcing bias. The way Facebook was used to influence the US election highlights other dangers; how ‘false news’ can be targeted at specific demographics.
2. Augmented Intelligence
Think of the ‘A’ as standing for ‘augmented’ and AI immediately becomes more of a positive than a negative. Augmented intelligence is about helping people work faster, smarter and safer. Perhaps the best example is in contact centres or on interactive websites where chatbots use NLP to handle routine customer queries.
Rather than replacing human agents, virtual agents handle first response ‘triaging’, giving people time to make the most of machine learning insights to resolve more complex queries – more like ‘cobots’ than chatbots.
3. Smart cars in smart cities
Advanced forms of cruise control have already brought AI to cars. But the big leap forward is expected to be autonomous, self-driving vehicles. There are many challenges to overcome to make them safe and efficient, like dirty sensors and life-threatening system failures, but interim technology advances already offer plenty of other possibilities.
Smart vehicles moving through highly networked smart cities will give us a rich source of real time data that can be used to improve traffic flows, pollution levels and road safety. At BT, we’re already in this market in the UK with Auto Mate, a mobile device that provides real-time driver, vehicle health and tracking data.
4. Explainable AI
With GDPR enforcing an individual’s right to find out how their personal data is used, explainable AI has become a requirement. If algorithms and machine learning are used to calculate someone’s credit rating, for example, the person has the right to know how the decision was arrived at – that means understanding how algorithms arrive at a decision. Not easy.
This brings in wider ethical issues. There’s a risk of bias in the most basic AI services because the people who write the models for algorithm-informed decision-making inevitably bring their own biases to the code. Human intervention will be needed to provide constant checks and balances.
5. Enhanced cybersecurity
IT security teams battling increasingly sophisticated cyber attacks have to identify real threats in reams of data generated by monitoring systems. By recognising a baseline of normal activity, AI and machine learning can help spot anomalies in patterns and send out early warning alerts.
We’re already using machine learning and deep learning algorithms to detect anomalies on the network. They’re surfaced on dashboards for our human experts to follow up and neutralise. Once again, it’s about humans and machines working together to solve problems.
Security is a big area of AI for BT, along with its role in smart cities. We’ve also contributed to the debate around ethics as part of a UK government consultation on the formation of a national Centre for Data Ethics and Innovation. It’s still early days but these are issues that we’d encourage every organisation to explore as the footprint of AI and machine learning inevitably grows.