Ayesha Khanna: Business reimagined with AI
AI can and will change the way we live and work - and we should prepare for that now. A liveblog of Ayesha Khanna at NEXT18.
Dr. Ayesha Khanna is Co-Founder and CEO of ADDO AI, an artificial intelligence (AI) advisory firm and incubator. She has been a strategic advisor on artificial intelligence, smart cities and fintech.
If you step out of the most advanced, developed countries, you find more excitement than fear about AI. Why? They think it offers them the opportunity to leapfrog other, more established economies.
Has the race already been won by China? China has invested so much in AI, that some say it could take the lead – and it has a three year plan from the chips to the high school books, to the police who have AI-enabled facial recognition goggles.
More and more countries are getting an AI strategy. Germany has one. France’s President Macron talks about putting humanity and ethics at its core.
One vision is the Black Panther version: “I saw the movie, it looks awesome”. Or maybe they are inspired by a Go champion beaten by an AI. The fourth industrial revolution is — or will be — AI. But people in the UK are sceptical, or puzzled. We can give away our agency by watching new technologies develop like a sci-fi movie.
AI is great at personalisation – it watches people’s micro-behaviour and gives them the right information at the right time. Facebook does this. Google does this – every one of us will get different search results. Netflix uses personalisation all the time. They mine data to make better and better recommendations – both saving them money and generating new business. 35% of Amazon’s sales come from the recommendations at the bottom. In China, the one day TMall sale generated billions of personalised pages.
In Dubai, they looked at all the citizen interactions with government, and integrated in one digital place – but engagement was still low. Why? They were treating every citizen the same way. They’re used to getting relevant, personalised services. Now, the government has a lot of data – and they can get more, and put it in an AI engine.
Busy mothers want to know about financial support for education, or yoga classes. Young entrepreneurs want to know about immigration or taxation laws. Our prototype showed them the information that mattered most to users, when it was relevant to them. Most AI work starts with small prototypes.
AI is coming to customer service, via autonomous agents – and by 2020 5% of all customer service transactions may well be handled without human intervention. Forgetting for a while about the human agents losing their job, let’s focus on the consumer – they want their problem resolved as quickly and simply as possible. Can AI extract meaning from torrents of emails? Yes – you can mine the key data, and run a range of AI models on it – and it can understand about 85% of emails.
Customer Service via text or chatbots is becoming old hat. People would rather just speak. Speech to text recognition is the new frontier, and in Asia, localisation matters, because there are hundreds of languages. Others are exploring the use of AI to recognise emotional state of the customer via facial recognition. Many AI experts are not comfortable with this – but it is out there.
It’s easy to say “we’re customer centric”, but most of us are egocentric, and are, at best, company centric. Uber and Grab are creating great pressure on traditional transport businesses, by taking their customers. To compete, traditional transit needs to think about the customer’s stories not their own. Why do people take cars in parts of Asia? It’s so hot there – bikes or public transport are too hot, and they don’t lead to their users arriving at work in pleasant condition.
To persuade them, you need a single app for inter-modal transit. Allow people to book and pay in one place, that allows them to customise the app with their goals – save money or get more healthy by walking further. To do that, you need to integrate all the data assets you need, and then customise the offering for users with AI. MobilityX is doing that in Singapore. The company can do things like suggest to companies that the need more scooters in some venues based on predicted trends.
This is using AI to become more customer-centric for users.
80% of Asia’s food supply chain comes from marginal farmers who live on $1 to $4 per day. And they’re incredibly vulnerable to weather or pests impacting their crops. Can we give these farmers insurance? The insurance companies think it’s too labour intensive to verify claims. How about using remote sensing data – the eye in the sky? Using satellite data you can detect pests via thermal imaging, or measure the amount of the crop destroyed. You can even prevent claims by pre-warning farmers so they can spray their crops.
This is AI making new sets of customers viable.
The AI redundant
Singapore is a small, new country, with no resources other than its people. What if jobs and skill have an expiry date? Tech could destroy between 30 to 50% of staff in banking. That could devastate a financial capital like Singapore. On Wall Street there are whole departments devoted to predicting the outcomes of likely disasters. Kensho is using AI to predict, for example, which cement stocks will go up in the aftermath of a disaster.
Singapore has decided to encourage AI, as it’s coming anyway. But, as they give grants to AI, they’re also adding education for potentially redundant bankers. They’re teaching AI to the bankers, so they know enough to co-operate with AI specialists. Any AI project team needs AI experts and domain experts, and the more they understand the key terms, the better that collaboration is.
Singapore is offering its citizens $500 to take any course, to get them in the habit of life-long learning.
Embedding values in our AI
There’s a great deal of mistrust based on the revelations of what Cambridge Analytica did with Facebook data. And there is a danger of bias. Google’s image recognition was trained on white males, and ended up categorising dark skinned people as gorillas. An HR AI was picking only men for interview – not because they were better candidates, but because it had learned from the company’s history of mainly employing men.
Everyone says Facebook is bad – but when they added privacy controls, their stock price dipped, because investors thought the newsfeed would become less useful to them. We need to stand up for our values in AI. Europe is good at that, and Asia is learning from you.
If we can build trust into AI, we can build better businesses.