Breaking News

Ask the Founders: What is Artificial Intelligence exactly? – WebWire

In the past few years, Artificial Intelligence has become more and more ubiquitous in our everyday tech, and it shows no signs of going away, with investment into UK AI reaching £2.42bn in 2019. According to the Tech Nation Report 2020, looking at data from 2015-2019, the UK is third in the world for levels of AI investment, second by deal count, and the only country of the top 5 AI nations to have demonstrated consistent positive year-on-year growth for the last 5 years.

A regular feature of sci-fi and sometimes horror, AI has often been forewarned as something to be afraid of, and while the ethical implications are something that need to be kept central to the development of AI, it’s not something that necessarily spells the end of days.

So what exactly is Artificial Intelligence, what does it mean for the consumer and where do we expect it to go in the next few years? We asked some members of the public for their main concerns surrounding the technology, and asked some founders from our Applied AI 1.0 programme if they could clear up some of myths, misunderstandings and concerns.

Mohammad Rashid Khan, Co-founder & CEO of Calipsa and Jamie Potter, Co-founder & CEO at Flexciton gave us their insights. 

What actually is AI?

What actually is AI? – Amy, Student

Jamie Potter, Co-founder & CEO of Flexciton  “John McCarthy, an American computer scientist, first coined the term ‘artificial intelligence’ in 1955. He organised the famous Dartmouth Conference in the Summer of 1956, which started AI as a specific field. At that time it was an area of interdisciplinary computer science research.

There are continuous debates amongst scientists of different disciplines, in addition to tech giants, on the standard definition of AI. We’ve noticed that the most common understanding of Artificial Intelligence is a machine’s ability to learn and improve from experience (Machine Learning). This definition, however, is only a snapshot of what we know AI is capable of.

We would like to offer our own simple, yet inclusive definition of Artificial Intelligence: An area of computer science where mathematical models and algorithms are developed and applied to simulate intelligent computer behaviours.

Algorithms often play a very important part in the structure of artificial intelligence. Simple algorithms are used in simple applications, whilst the more complex, help frame strong artificial intelligence. 

AI applications cover a very wide spectrum – from ‘simple’ Machine Learning tasks, to very complex problems-solving algorithms. As consumers, we are surrounded by AI. Some help us with the most basic of tasks, for example voice dictation – a virtual assistant that uses voice recognition to answer our questions, or a saving app that records and learns from our spending behaviours and automatically sets money aside, to the more complex applications like autonomous vehicles.

AI is also widely used in the manufacturing industry. For example, at Flexciton we have developed advanced mathematical optimisation techniques to help factories perform production in the most efficient way. Every day, in any factory throughout the world, someone has to decide what to manufacture, in what quantity and order. With only one hundred different products to make there are more options to choose from than atoms in the entire universe. Flexciton’s AI-powered technology is capable of selecting the most optimal production scenario from all possible ones”

Will AI replace human interaction completely?

From a customer’s point of view, will AI replace human interaction completely? – Judith, Civil Servant

Mohammad Rashid Khan, Co-founder & CEO of Calipsa “We believe customers shouldn’t be concerned about AI replacing human interaction. There are a lot of misconceptions out there about what AI is, what it can currently do and how it’s used. Forbes published a very good article which helpfully splits AI into three categories: Acceleration, Augmentation and Automation. 

Acceleration uses AI to speed up human processes, such as the autofill function on an online form. 

Augmentation accelerates a human process, but also helps to improve the quality of the final product; for example, Netflix’s recommendations for similar shows you might enjoy. 

Automation achieves tasks without any human interaction or input. In reality most true automation is still being developed, such as driverless cars. 

At the moment, much of the AI that helps us in our daily lives falls into the first two categories, so that we can carry out activities faster and better. It’s important to consider the context, too – not all activities will benefit from becoming fully automated, since human interaction and reasoning is much more useful in plenty of situations. Healthcare is a good example of an industry where AI and humans both have important roles to play. 

Even as AI becomes more sophisticated, there is still plenty of room for human interaction”

Jamie Potter “We are living in a complex world where expectation for higher performance is in constant  demand. This requires a different approach. At Flexciton, we believe AI can help us achieve greater results quicker and in doing so, make things much easier for humans.

While some areas may become much more automated, in the future , demand for new skills and space for humans to contribute will become apparent. A good example is in the fintech sector with neobanks offering a highly automated customer interface without the need for physical branches. This created a need for highly skilled engineers and developers and a customer service team capable of delivering quality service over the phone or online chat.

Flexciton works with various industries, and we can see how our AI-powered technology can help to  gain efficiencies which allow more time for people working on a shop floor to deal with more strategic questions.”

How does it benefit the consumer?

In the same vein, I assume AI is efficient and cost effective from a business point of view – but how is it beneficial to the consumer?  – Judith, Civil Servant

Mohammad Rashid Khan “Reducing the time people spend on time-consuming and repetitive tasks passes plenty of benefits on to the consumer, as well as saving businesses money. For instance, at Calipsa we prevent more than 30 burglaries every month thanks to our False Alarm Filtering Platform.

We are training our AI to save security staff time and money that would otherwise be wasted on checking false alarms sent by CCTV cameras. Now security businesses can focus on the alarms that matter, rather than trawling through lots of false ones. As well as saving money and resources, being able to effectively deal with true alarms and fight crime can help to boost morale for both the security staff and their clients. 

It would be impossible to achieve this level of safety through human efforts alone, simply because we don’t have the time or energy to check security cameras as consistently as AI can. People eventually get tired, and it is difficult to carefully monitor multiple camera feeds at the same time. Calipsa’s technology helps security staff to do their jobs better, which in turn makes the world a safer place.”

Jamie Potter “AI applications bring countless benefits for consumers – both directly and indirectly. For example, Flexciton technology can bring significant efficiency gains in the manufacturing process, which will inevitably lower operating costs. These savings can be passed onto consumers with lower product prices.

Many companies use AI to improve customer experience by determining the best customer-product match. Netflix, for instance, provides highly accurate predictive technology based on customer’s reactions to films. It analyses billions of records to suggest films that you might like based on your previous reactions and choices of films.

When used in healthcare, Artificial Intelligence can bring critical benefits by eliminating human error or improving diagnostic accuracy. The recent research study by Imperial College London and Google Health shows DeepMind’s medical AI system can outperform doctors on identifying breast cancer from X-ray images.

Most of the benefits of AI applications in use today would fall into one of the following categories: personalisation of products/services, convenience, accessibility, saving time and money.”

Where can we expect to see AI in the future?

What are the potential applications of AI and how soon will it be widespread? – Hafi, GP

Mohammad Rashid Khan “AI has the potential to be applied in all kinds of areas, though how widespread each particular type of AI becomes will vary from case to case. Areas adopting AI include healthcare, finance, insurance, entertainment and, in Calipsa’s case, security. 

One kind of automated AI that is already causing concern is driverless cars, as they could potentially put thousands of drivers out of work. However, as we’ve already seen, the majority of AI currently out there requires some level of human input for it to be useful and effective. 

In a paper published jointly by AI experts from Oxford University, AI Impacts and Yale University, they discuss the results of a survey conducted to find out how long participants thought it would take to achieve high level machine intelligence (HLMI); that is, “when unaided machines can accomplish every task better and more cheaply than human workers.” 

The survey’s participants were mostly other experts in the field, and responses varied widely. It seems that even among experts, it’s extremely difficult to agree on exactly when and how AI will become widespread! 

For now, it looks as though AI certainly is on the rise in many areas of our lives, and when it’s used appropriately, it’s not something to be afraid of: it is there to provide solutions to problems.”

Who is accountable for AI when it goes wrong?

If AI used for purposes such as diagnostics or security makes mistakes or does harm, who would then be responsible? – Ayhan, Radiologist

Mohammad Rashid Khan “At Calipsa, we think that the companies building these technologies should take liability and responsibility for any mistakes made by their AI. If our security AI system makes a mistake, which has consequences for one of our security clients, then we take the liability for it. 

However, statistically we’ve found that the level of AI error is lower than human error. On the whole, when AI is developed to carry out a particular task more effectively than a human can, its capacity for error tends to be much lower – particularly because the humans training the AI system have the capacity and reasoning to detect its mistakes and correct them.”

Applications to join Applied AI 2.0 are currently open! Find out more.