Artificial Intelligence (AI) is a term that is readily recognisable but often misunderstood. It is frequently misused by the media, the hype around potential and expectations is huge, and as a result, rarely enables business owners the ability to see how true AI can be applied. Business expectations are not aligned to the adoption of AI and this means that many firms are not benefitting from the exceptional possibilities that technological advancements can offer right now. It is important not to be drowned in the promise of the future by sacrificing the potential of the present day.
“We always overestimate the change that will occur in the next two years and underestimate the change that will occur in the next ten. Don’t Let yourself be lulled into inaction.” — Bill Gates
AI Hype
AI is often viewed as the tool that will revolutionise a business’s entire operations, or that it will take over and render human workers unnecessary. Lloyd’s of London, for example, recently detailed that an AI partnership promises that in a decade a significant part of the insurance industry will be powered by AI.
The BBC recently reported on the use of AI by a Welsh company to detect bio-weapons in North Korea, along with a number of other news inclusions of similar stories. However, some of the industry’s leading names are quick to downplay the hype, with Google’s AI head, John Giannandrea diffusing the issue by stating:
“There’s a huge amount of unwarranted hype around AI right now… This leap into, ‘Somebody is going to produce a superhuman intelligence and then there’s going to be all these ethical issues’ is unwarranted and borderline irresponsible.”
It seems that there is a misunderstanding about the capability of AI — what it can do now and what we will have to leave on the wish list. It is a vicious circle because the media continue to play on the hazy AI definitions, which contributes to the hype and in turn generates more confusion.
AI Reality
Will all of our jobs be automated away by bots? In short — no. We believe the technology has incredible potential to impact all aspects of our lives. In fact, a study by research experts at the Pew Research Centersuggests that advances in robotics and AI will create more jobs than they displace over the next decade. The study demonstrates the fact that historically, technological advancements have always created more jobs than they have destroyed.
Some of the hype surrounding AI is thanks to Hollywood but in reality, today’s AI is not sci-fi and we are not realistically about to adopt autonomous R2D2s into our daily lives. What is realistic for today falls in the category of narrow AI — “sci-non-fi”.
With narrow AI, there are a variety of techniques:
1) Functional programming: This is effectively what we all recognise to be programming — The act of demanding a result from a computer or a way to tell a computer to do something deterministically.
2) Machine Learning (ML): A more advanced form of programming where we provide the computer with the data and ability to learn for itself through coded rules (algorithms). The computer will be able to offer probable predictions on an outcome and with a greater input of data, the ability to produce accurate predictions will increase.
3) Deep Learning (DL): This more modern version of machine learning is often the basis of media hype. Unlike machine learning, deep learning enables the computer to learn simply by giving it examples as opposed to machine learning, where the computer is asked to find specific information. Deep learning is therefore much more like a human brain and is particularly effective with things like image recognition apps. If you have an iPhone you will notice the Photos app now has facial recognition.
4) Deep Reinforcement Learning (DRL): The further evolution of machine learning is DRL, which is able to combine deep learning and reinforcement learning (the learning through experience and trial-and-error, which is reinforced by punishment or reward). This type of learning mirrors the ways in which a child develops which, is the notion of learning by trial-and-error, solely from rewards or punishments. DRL mimics how children learn — they observe other people doing things, they try things out and depending on the reward, they either repeat them or not.
Machine learning technologies are more available than they ever have been and this goes some way to explaining the increasing media hype around this space. The developments have been led by advancements in the following areas:
1) Machine learning algorithms run by developed infrastructure. There have been huge technological improvements in server-processing abilities, storage capabilities and cloud accessibility.
2) New algorithms are being developed more quickly and this makes them more accessible.
3) Greater awareness and interest in data scientists. Data is the lynchpin to developments in this area and with greater availability coupled with innovations in algorithms and data scientists’ increased profiles, the field is propelled for further advancements.
The Real-World Application Of Machine Learning
We recently worked on a machine-learning project for Hackney Council, using data science to predict tenants who are most likely to fall into rent arrears. Digital Transformation Manager for Hackney Council, Tom Harrison, explained the relevance of data science to local housing:
“What if we can predict those most at risk of falling into arrears so that interventions can be targeted to prevent the problem before it occurs? Within the council we have a Financial Inclusion team that helps residents with financial planning and can point them towards training to help them get better paying jobs. If we could better target that team’s resources to those most at risk then this wouldn’t just help reduce rent arrears, but give our tenants more control of their finances and help tackle unsecured debts, or payday loans.”.
Working with Pivigo, Hackney Council will strive to highlight vulnerable tenants through the analysis of data and the ability to incorporate specific parameters “such as anticipated inflation or wage growth rates to see how this may impact on our residents in given scenarios”. As an additional benefit to the council, personal data is not required in order to develop the model and so personally identifiable information remains secure and will only be relevant once vulnerabilities are detected and only to those departments who can intervene to help. This is a really useful application of machine learning that benefits not only residents at risk of falling into financial difficulties, but also help the council budget spend more efficiently.
Conclusion
AI will permeate all businesses in the next ten years and inevitably bring revolutionary changes. There are enormous time, money, and manpower savings to be made by automating simple processes; and as the technology becomes more advanced the use cases will become more exciting and closer to the hype of today. But there are applications of Narrow AI today that can give businesses considerable advances in increasing productivity, efficiency and cost savings.
The hype surrounding AI is extensive, and there are many reasons to be excited by this. The difficulty can be cutting through the enthusiastic visions for the future, to understand the tangible and realistic value AI can unlock in your business today. Businesses who understand the basics can quickly become passionate about the future with the confidence that automation is not the end for human superiority in the workplace.
Comments