Unless it is fed on high quality data, AI will be delayed on one of its most vital missions – to earn our trust
Estimated reading time: 5 minutes
We could all benefit from artificial intelligence (AI) at the moment. Right now, supercomputers across the world are analysing petabytes of medical, agricultural and climate change data in a bid to save us all from disease, pestilence and floods.
In the haste to find solutions to our challenges, it is often overlooked that the quality of machine learning and neural networks is directly related to the integrity of the data they ‘feed’ on. So these Automations mustn’t be fed useless intelligence. A developing Automation must be fed data of the purest quality if we are to trust the judgements coming down from AI.
For this very reason Louise Wright, Head of Data Science at Britain’s National Physical Laboratories, is running a campaign to raise awareness of the need for healthy, clean data.
As with any diet, the provenance of the data is a major concern. Portion control is vital too. “We need to have more confidence in how data is collected,” says Wright, “The problem is that sensor prices have fallen and everyone is a data collector now. There’s too much information.”
It’s important not to avoid the fundamental question, says Anis Chemli, VP of Sales and Marketing at Guavus, a Thales company and pioneer in AI-driven analytics for communications service providers (CSPs). The big question is: do people actually understand data? Intellectually yes but operationally no, according to Chemli. Guavus helps CSP clients to think about what they need from their subscriber, network operations and device data, and apply analytics to get new insights on their customer experience and operations in real-time. Since data is their most precious asset, they need to focus on the fundamental principles - the use case, the source of the data and its quality.
“It’s all about what you need from that data,” says Chemli, “For that reason, we often advise the vendors of the IoT devices to change their systems to provide the right data.” Even if there is no machine learning involved, the wrong kind of data won’t add anything to the sum of knowledge, in which case it will only detract.
Context adds value to data and that superior knowledge lessens the perceived risk, according to Charlotte Gribben, a partner multinational service provider Deloitte who has specialized in analysing Digital Risk for 15 years.
Gribben makes the tough calls on every new technology option faced by her clients. Gribben has their trust because it takes courage and integrity to hold out when everyone’s heads are being spun by aggressive marketing. ‘The most dramatic sign of maturity is that it cuts across all areas of the business now, from accounts to corporate communications through marketing to operations,’ says Gribben.
IoT technology can improve world’s sustainability by saving energy in the long run. Smarter objects mean more efficiency. Collecting, analyzing, and measuring the behavioral aspects of IoT devices will enable societies to fine-tune their energy consumption and can be used to reduce the impact on the environment.
In the IoT world, data matters now more than ever, says Chemli at Guavus.
The digital future, as represented by the Internet of Things (IoT), requires something of a leap of faith and AI’s pioneers such as Guavus, Deloitte and NPL stress the importance of earning the trust of the public who are being asked to live in a new world regulated by machines which have yet to develop empathy.
Related content: The flow of data - sharing information responsibly
New Podcast series: This is Artificial Intelligence