The Yet Untapped Potential of AI

As the uses and potential of machine learning grows, the V4 countries are uniquely qualified to expand in this specialised area of the market

Jerzy Brodzikowski
15 March 2017

The moment of the next revolution in the history of mankind is fast approaching – the difference between this and all previous revolutions, both social and economic, is that this time we won’t be in it alone. We will be accompanied by complex computer systems carrying out activities, which, due to our cognitive abilities, were previously reserved only for human beings. The era of artificial intelligence is here.

Probably the most famous examples of the application of artificial intelligence, though rather simplistic, were the chess duels between grandmaster Garry Kasparov and IBM computer Deep Blue in 1996 and 1997. These duels were supposed to show that a machine can not only perform certain actions similar to a human being, but also outperform its human counterpart. The first match in 1996 ended in victory for Kasparov. The second, which took place a year later, forced the grandmaster to recognise the superiority of the modernised computer.

It’s been almost twenty years since the aforementioned historic duel, and artificial intelligence, which seemed so abstract back then, has stealthily penetrated our daily lives; it has changed our relationship with our surroundings, continually transforming our environment into a more and more digitalised world. To better understand its level of penetration in our society, it is worthwhile to examine what is currently the most developing segment of artificial intelligence – machine learning.

As the name suggests, machine learning consists of creating computer systems which can learn first in a simpler form with human assistance, next in a more advanced form (i.e. with less human assistance) ultimately reaching instructional autonomy. The solutions of the first kind are already available on a mass scale – for example, in online stores, which learn the behaviours of their customers and suggest other products (e.g. in in the form of an advertising banner) which could potentially be of interest in them. More complex machine learning systems include speech recognition, available in personal assistants such as Google Now and Siri on mobile operating platforms like Android or iOS, or facial recognition on photos through online services such as Flickr, Google Photos or Facebook. Image analysis is especially interesting due to the fact that it uses machine learning that can be defined as deep learning, which is structurally inspired by human brain neural connections. Deep learning is currently considered as the segment of artificial intelligence with the greatest potential, and it can be applied to automatic translation (interpretation) of texts, colourisation of black and white photography and videos or digital graphological analysis.

Unfortunately, most of the above solutions require constant collaboration between large teams of scientists and programmers which improve the algorithms, in order to increase their models’ efficiency in achieving a specified objective; this might include creating customer behavioural patterns on an e-commerce website, a correct diagnosis of a voice command or an accurate selection of a friends face on a photo. As in the entire science of artificial intelligence, having well-defined goals in machine learning is paramount, while the biggest barriers are the lack of high quality data and the still limited computing capabilities of processors.

In the more advanced area of machine learning – which is now being developed by major players in the world of technology, such as Google (DeepMind and Google Brain) or Facebook (Facebook AI Research) – more complex models of AI are being created. These systems, when properly programmed, teach themselves how to reach a certain goal. The best example of such an application of machine learning can be found in a London-based company DeepMind (bought in 2014 by Google), which creates its artificial intelligence systems so that they learn how to play computer games from the 1980s. When playing a simple game which involves shooting-down of alien creatures from a spaceship, the system developed by DeepMind has one goal: survive in the virtual space as long as possible. The task for the system is, in this case, to learn to survive by repeatedly attempting the game through the method of trial and error. It is this ability to teach and improve its own operations (although, for now, with human assistance) which will cause the speed of technological development in the future to gain even greater momentum. Imagine a normative, contemporary system – which is supported by a large team of programmers working on enormous databases and spending hundreds of hours to correct the AI model – being replaced by a system which learns to correct itself in a shorter time with greater efficiency.

The potential for the development of the AI sector is immense, and its global value can reach $36 billion by 2025. However, it is true that the greater part of this market will be taken by the largest technological companies such as Google, Facebook or Microsoft, which, for many years, systematically invested in the development of artificial intelligence, and the results of their work can already be witnessed in the services that they offer. Fortunately, there are smaller segments of the machine learning market – usually focused on specific, individual solutions – which are being developed in the CEE region (most notably in Poland) by technological start-ups.

Examples of such companies are Craftinity, Voicelab, Growbots or Deep.BI. Craftinity specialized in analysing and learning of hierarchical databases using the deep learning method with which they help detect systemic anomalies and provide semantic analysis for their customers. Voicelab deals with speech recognition and speech analysis. The founders believe that in the future they can create a system of voice recognition that will work without the need of an Internet connection, which is today required by Google Now and Siri. Growbots developed an innovative system for data analysis which helps retailers search for customers. Thanks to their system, e-mail sales campaigns for their clients are automated and the sales people or product owners can focus on getting in touch with the client who has a higher probability of being interested in the product. Deep.BI, on the other hand, measures raw data about online publications, such as readers’ behaviour and content performance, analyses them in real time which allows for a better understanding of the needs of customers by content creators and publishers, at the same time enabling them to increase their audience.

The biggest problem faced by companies currently operating in the segment of artificial intelligence is funding, which in this part of the world is considerably smaller than, for example, in the UK and the United States, where private equity funds are driving the development of experimental technology. A great opportunity for the sector of machine learning in Poland is a government run program Start in Poland, which, through a venture capital fund PFR Ventures, will offer 2.8 billion PLN for the development to innovative start-ups. Alternatively, Horizon 2020, a programme run by the European Commission has earmarked 38 million for research on big data, which includes machine learning. Unfortunately, even the aforementioned amounts, pale in comparison with the money that is pumped into research on artificial intelligence in other places.

What is also worth mentioning is the huge programming potential which has been dormant in the CEE region and could be used in the development of artificial intelligence technologies. The most notable potential is in the fact that young representatives from countries comprising the Visegrad Group have been, for many years, achieving enormous successes at the International Olympiad in Informatics (IOI); in the historic “hall of fame”, three countries of the V4 group are in Top 20 (Poland has a high second place in that ranking). Unfortunately, these young programming talents are usually fished out by the great players from the West. The best example of this is the recent employment of Filip Wolski, the most successful Polish Informatics Olympian, by California based Open.AI, which is financed by none other than Elon Musk. Interestingly, one of the founders of Open.AI, which aims to maximise the potential of open artificial intelligence is also a Pole, Wojciech Zaremba.

The sector of artificial intelligence has a very bright future and is therefore worth being supported at both the research and business levels. With so many dynamic, technology-dependent changes in store for humanity in the near future, there is ample space within the machine learning sector which could be allocated not only by people from the CEE region working for foreign companies, but also for companies from the CEE region. These companies can utilise the dormant programming potential and pervasive hunger of this region to build an economy similar to those in the Silicon Valley. It is worth retaining those talents here; after all, there is no better resource in business than the human resource and, in particular, one that creates artificial intelligence.

Jerzy Brodzikowski is Community Development Manager at TechHub Warsaw, where he helps over 40 technological startups develop to their maximum potential. He specializes in new technologies and their application in professional life, startup development, public relations, social media, and project management.