Interviewer: The Swedish Chamber of Commerce for the UK
Interviewee: Tommy Högström, partner at Applai
What exactly is Artificial Intelligence (AI) and what’s the difference from Machine Learning (ML)?
Although its many definitions I would say that Artificial Intelligence (AI) is an umbrella term for technologies that enable machines to mimic human intelligence, such as seeing, comprehending and learning (Machine Learning). In other words, ML is a sub-technology of AI that learns from data, often to predict the future of different kind of human or machine behavior. Actually, almost all the economic value created by AI today is through supervised ML, analyzing input to determine an output with X percent accuracy (for example predicting with 95% accuracy that that a customer will churn based on his or her behavioral pattern).
How far have we come in the development of Artificial Intelligence (AI)?
Far enough to transform the way companies are run, but far away from robots taking over the world. It’s estimated that 50% of all human performed tasks can technically be done by machines, today. With that said, far from all tasks needs AI, some can be done with simple automation.
Where can we find Artificial Intelligence (AI) in our everyday lives?
AI is already all around us, enabling Apple’s Siri to understand you, email clients to detect spam and retailers to recommend products based on your internet behavior. Self-driving cars is on the rise and digital assistants are rapidly becoming harder to distinguish from humans. What AI really brings to the table in our everyday lives is convenience and soon most transportation and customer support will be done by machines.
How are companies using Artificial Intelligence (AI) today?
The application areas where companies are using AI today are many and quickly increasing, however discussions of what data companies should be allowed to use is still a moral dilemma. For example, should they be allowed to base individual insurance pricing on criminal history or geographic origin?
- Customer Intelligence – AI enables companies to know what consumers want, when they want it and how they want it, but also exactly what customers are monetarily worth. Most large retailers have already started to analyse their customers and to personalize communication, however increased customer demands for personalized communication will ramp up this development during the upcoming years.
- Financial analytics – Various type of transactional data in combination with customer profiles, industry trends and text data are of tremendous value to banks and insurance companies, using it to perform advanced risk calculations and fraud predictions. For example, insurance companies want to know the exact risk of a customer defaulting on a loan, to maximize price and margins.
- Manufacturing – Producing companies collect vast amount of production data such as equipment usage, vibrations and temperature which can be used to maximize production up-time and optimize quality but also to predict maintenance needs.
- Supply chain – Using smart algorithms companies can perform precise availability forecasts of products in stock and delivery times to customers.
- Intelligent automation – Commonly applied areas for automation are chatbots interacting with customers and various manual data entry task such as order processing or internal IT ticketing.
How can the world benefit from AI?
AI have the ability to solve the largest problems of our time, such as solving climate change through optimizing resources and to predict and prevent diseases such as cancer. Other areas where AI can be used to benefit humans are automated transportation, robot friends (imagine R2D2 – amazing), improved elder care and improvement of our physical bodies (cyborg technology).
How do businesses get started with AI?
- Start by identifying use cases, i.e how Artificial Intelligence can drive value in your company. Think about current challenges or opportunities and ask an expert if you are not sure how/if AI can solve your challenges.
- Next step is to assess the feasibility in terms of data access, quality & time horizon as well as relevant business processes. If the right data is not yet accessible, it’s crucial to start thinking about data structure and acquisition.
- The most important step in implementing AI is making sure the solutions brings real business value, therefor you need to estimate potential value, costs & risk, per use case. From this analysis you can create a strategic roadmap towards 100% smartness, focusing on what will bring your organisation the most value.
- Now it’s time to start building models with a ‘quick value’ mentality. Bring in experts and set up a pilot project to iteratively build a value creating model and make sure to learn in the process.
What does the future look like for AI? What can we expect?
While digitalization has made us addicted to screens, AI has the ability to detach us from them, to communicate through technology in a whole new way through voice and movements. I believe the lives we live today will be considered stone age in a near future.
In terms of jobs leading research points to more jobs are to be created (than roboticized) due to AI in the foreseeable future. Here are some interesting statistics:
- By 2020, AI is estimated (according to Gartner) to generate 2.3 million jobs, exceeding the 1.8 million that it will remove.
- Experts (from Yale and Oxford universities) say there’s a 50% chance AI will outperform humans in every job in 45 years (and 10% chance that it would happen in 9 years.). Whether we let AI do them is a whole other question.
While AI is promising to society, huge risks are associated with this development including AI warfare, built in discrimination and loosing control over what AI do. I wouldn’t be surprised to see AI as the next big election issue.
That Artificial Intelligence will transform companies as we know them is for sure, but whether AI will be the last human invention is for the future to tell.