Published 19. Apr. 2017
Artificial Intelligence is Becoming a Part of Everyday Life
IoT product development challenges existing R&D thinking and enables more impactful product development.
by Markku Rimpiläinen
How can you know whether you are dealing with an artificial intelligence system or a human being on your bank’s customer service chat? In fact, you cannot. Artificial Intelligence systems already understand natural language to the extent that they can reply to customers’ questions, provided that the subject area is restricted.
Artificial Intelligence also possesses other fine properties for customer service. It can adapt to the behaviour of the customer. It learns the best ways to serve from its human colleagues and knows in real time what kind of problems the customer has had. In addition, Artificial Intelligence can assess when the answer it has found is no longer reliable enough. At that point, it will forward the processing of the problem to a human.
In addition to customer service, Artificial Intelligence is already present in many other everyday applications.
The latest version of Google Translate deduces the most accurate translation for whole sentences, no longer for individual words. In medicine, systems based on Artificial Intelligence reliably diagnose severe cases of cancer and find the best treatments for them. The Internet of Things and Artificial Intelligence are developing side by side; in the future, Artificial Intelligence will be embedded and unnoticeable.
Still, there is nothing spectacular about Artificial Intelligence, according to Tuomo Haukkovaara, Vice President, Solutions, IBM Nordic.
“An Artificial Intelligence system can perceive things and combine them. It learns to understand information and draw conclusions based on it. A human being acts in quite the same way,” Haukkovaara says.
An Empty Artificial Intelligence System Cannot Do Anything
Artificial Intelligence is significantly different from older IT systems. It understands, interprets, and learns.
However, an Artificial Intelligence system does not learn anything by itself; it needs to be taught.
First, a so-called basic corpus is created for the Artificial Intelligence, which means that the basic information needed for problem-solving is entered into it.
“Without a basic corpus, an Artificial Intelligence system does not know anything and cannot do anything; neither can it operate outside its own area of experience. When the machine is used and feedback is provided to it, it will begin to combine things and draw the correct conclusions.
In time, the correct answers will be found even for complex questions. At that point, it becomes apparent that this is a very different thing from anything we have had before,” Haukkovaara says.
A Smart Helper
What is Artificial Intelligence good for, then?
“In many applications, Artificial Intelligence analyses information and then offers a human expert alternative solutions based on which he or she can make a decision. This is typically the procedure, for instance, in the field of medicine.”
Another field of application that is rapidly becoming commonplace is image recognition. Artificial Intelligence can interpret various things from an image.
“Artificial Intelligence can assist insurance companies in assessing motor vehicle damage. It can tell the vehicle’s model and the severity of the damage from pictures, as well as prepare a cost estimate for the repairs based on the available information.”
Artificial Intelligence systems can also combine information from different sources.
“Artificial Intelligence could estimate the future price development of a company’s shares by combining financial performance data, stock exchange releases, news items and discussions in the social media.”
An obvious field of application is the Internet of Things.
“Data communication is already nearly cost-free, and various sensors are very cheap. The Artificial Intelligence needed can be obtained from a cloud service. Machines are beginning to understand their own status and, at the same time, learning to serve people better.”
People are Predictable
What should a company do if it wishes to start using services that are based on Artificial Intelligence?
“First, you need to think which questions you want to have answered to ensure better success. Next, you need to investigate how you can utilize information that already exists.”
In many companies, Artificial Intelligence can be used to predict consumer behaviour.
“People are very predictable in the end,” Haukkovaara says.
Artificial Intelligence can analyse the personality of an online shop’s customer and make personalized offers based on the analysis.
In Germany, a chain of bakeries has observed that weather has an impact on the sales of pastries. Sweet pastries sell particularly well on a cloudy weather. If a bakery can estimate the local weather and the demand of produce correctly, there will be efficient sales with little waste.
“In a business with low margins, even small savings can make a great difference,” Haukkovaara says.
Cheap and Expensive
Many affordableArtificial Intelligence applications have already been created. For instance, IBM’s Watson has 40 ready-made functionalities, adapted to the world of the Internet. Watson works in a cloud service and is available for use when needed.
Some of Watson’s basic applications, such as image recognition and a personality test based on written text, can be tried out for free by anyone.
At the other end of the spectrum, there are wide-scale tailored solutions where an artificial intelligence system is built from scratch to solve an extensive set of problems.
Artificial Intelligence Needs Ethical Guidelines
Haukkovaara is certain that artificial intelligence will change the world. Genomics and artificial intelligence are causing medicine to develop at an incredible pace at the moment. Diagnoses are becoming more accurate, treatments are increasingly customised for the individual and cures are found even for rare diseases.
Haukkovaara also sees huge opportunities in education.
“Machine-guided and machine-assisted learning can bring knowledge to places where it is currently not available. For instance, there is an estimated shortage of 300 million teachers in Africa.”
Work is changing everywhere. Work is destroyed but also created. Haukkovaara is one of those who believe that alongside the traditional white collar and blue collar jobs, a new brand of “new collar” professions will appear. They will involve applying the latest information technology but will not require a long formal training.
Artificial Intelligence can also be used dishonestly. Because of this, Haukkovaara calls for ethical rules.
“In my opinion, users should be told which parts of a service employ Artificial Intelligence. In addition, it should be told how the machine has been taught and who has done it. This is already the procedure used in medicine.”
Dream of a Human-Like Computer
The idea of a human-like computer has existed for as long as the computers themselves. Haukkovaara came across the first applications already in 1985.
“At the time, we were talking about expert systems comparable to Artificial Intelligence. We thought that they could be used in engineering sciences for demanding calculation tasks. However, they were never taken in common use.”
Artificial intelligence appeared in headlines the next time when IBM’s Watson defeated human opponents in chess and the Jeopardy quiz.
“We were at a whole new level then. The power of computers was based on deduction and combining facts, no longer on raw calculation power.”
Haukkovaara is certain that the development will continue rapidly in the future as well.
“Production-related and commercial problems will be sorted out once science has first solved its problems,” he says.
Who is Tuomo Haukkavaara?
Vice President, Solutions, IBM Nordic
General Manager of IBM Finland in 2011–2017.
Symbio will be attending 600Minutes Innovation and Product Development in Finland on the 4th of April 2017
Chief Solution Executive, Symbio Finland