fbpx
Expert Interviews, Ki-commerce

"There is a danger that Europe will lag behind in terms of artificial intelligence."

June 2, 2019.
Europe lags behind in terms of artificial intelligence
Let's read the text
subscription

Required reading: 5 minutes

Why are Google Duplex or certain Alexa Skills only available with a delay in Germany? Are the English and Chinese speaking countries technologically dependent on us? Why should Germany and Europe invest more in companies that promote the development of artificial intelligence? Sarah Al-Hussaini, inspirational AI expert and one of the founders of Ultimate.ai, answers these and other questions about the Natural Language Processing and Development (NLP & NLG)

 

We've just seen a number of innovations at Google I/O based on natural language processing and generation. Why are these features first available in the US and UK and why does it take so long for them to be available in other languages?

Like most Internet giants, Google is an American company. Google's innovation decisions are made in the USA and are market-oriented. To serve its English-speaking core market and reduce complexity, Google's developments begin in natural language in English. This means two important things.

First, Google's natural language algorithms are developed and optimized for English conversations. Languages are very complex and diverse, so that the optimized algorithms are not necessarily applicable for different dictionaries.

Second, Google needs to collect huge datasets to develop innovations based on natural language. This is a very expensive and time-consuming process. When companies primarily develop intelligent systems for English, they need and create English data collections. There are therefore fewer sets in German, Spanish, Swedish or the over 4,500 other languages spoken in the world. The scaling of similar innovations is therefore a massive investment requirement and a major obstacle for the countries concerned. This is also the reason why in regions with on average fewer inhabitants (i.e. fewer buyers) - such as Scandinavia or the Benelux countries - many technical innovations in natural language cannot be found (e.g. Google Home is not available in all these markets).

This is of course risky for Europe. As intelligent systems are predominantly developed for the English language, a technological gap is created. Because available products and services for other languages are limited or less advanced. If this remains so, there is a risk that Europe will lag behind in terms of CI.

 

What are the consequences of delayed availability of word processing functions in non-English countries? What's the worst that can happen?

Innovation is built on innovation: English language systems are already smarter and can do more than intelligent assistants in other markets. If technology based on NLP or NLG and the available datasets continue to be developed primarily for the US, UK and Co, there is a risk that the gap will widen further.

New products and services in non-English speaking markets will therefore be less and less advanced. AI is a horizontal technology that is designed to change almost all industries. Failure to democratize the AI could therefore have a negative impact on everything from education to health, transport and growth.

In the worst case scenario, we as a society will become increasingly accustomed to living and working in certain languages as new innovations continue to be available only in English. In the long term, this threatens the multilingual diversity of our planet.

 

What can or should countries do to avoid these disadvantages? Can government investment or national AI strategies make a difference?

Government action can play a key role in the democratisation of CI. On the one hand, public funding can promote research and development of technologies in national languages. For example, grants, subsidies and tax benefits may be granted to companies that use and/or produce applied AI with natural language. This includes grants to young companies at the top of the innovation chain. Finally, governments in non-English-speaking markets can also assist in the collection of large databases. These may be publicly available and reduce barriers to the development of AI technology in their markets.

 

In your opinion, which country - USA, China, Germany - has the most promising AI strategy so far and why?

Today, most AI development is done by the largest Internet companies in the world. As the majority of these are located in China or the USA, these countries are far ahead. The Chinese government has also been very active in advancing its AI strategy, providing significant funding for AI research and business, and restricting the rules for data collection and use. In order to keep up, the USA and Germany must develop and commit to national AI strategies on a similar scale. This must be done quickly, as the technological gap between countries is widening.

 

You are one of the founders of Ultimate.ai - how do you use NLP and what distinguishes your technology from your competitors?

Ultimate.ai uses NLP to help some of the largest companies in Europe scale their customer service. Our AI technology supports the work of customer service representatives by providing high-quality, real-time response recommendations. We also automate repetitive customer service tasks and processes to avoid manual work and free up more time for more valuable activities.

Our main innovation is the use of applied deep learning to develop a solution that is completely language independent. Our NLP can be used in any language and is state of the art. This is a fundamental advantage for our customers, who are large European companies offering customer service throughout the continent.

Our innovation has its origin in our Finnish roots. When we founded our company in Finland in 2016, there was no open source natural language technology capable of processing the Finnish language. As a result, we built everything ourselves and used the latest breakthroughs in deep learning to develop our neural network architecture. Today, our technology is capable of processing large amounts of unstructured call data in any language and training an AI model that supports, automates and analyzes the work of a customer service representative.

We will continue to invest in multilingual NLP as we continue to scale within Europe. With over 130 million customer service representatives around the world working in hundreds of languages, we believe strong technology is needed to fundamentally change the market.

Sarah Al Hussaini

Sarah Al Hussaini

Sarah Al-Hussaini is Chief Operating Officer at Ultimate.ai and loves loudly your Twitter profile Technology, people, friendliness and questions that make them think. The founder studied economics at the University of Birmingham and was a scholarship holder specializing in the Indian and Chinese markets. Already at school Sarah won the UK Mathematics Challenge Gold Award and Best in School Award three times in a row. Earlier this year Forbes magazine named the inspiring AI expert in the popular "30 under 30" list.

Picture: Amir HamjaEyeEm.

These contributions should please you

No comment

Comment now