Artificial Intelligence: A Critical Overview Of Where We Are Today

October 9, 2017

I have been fascinated by the field of AI since my early days at university, where I remember looking through the list of courses on my Computer Science degree and reading: “Artificial Intelligence: 4th Year Course - 2 Semesters”.

Artificial Intelligence

I have been fascinated by the field of AI since my early days at university, where I remember looking through the list of courses on my Computer Science degree and reading: “Artificial Intelligence: 4th Year Course - 2 Semesters”.

This was back in 1990, when personal computers were scarce and expensive. The best that I could get my hands on was a ZX Spectrum or a Schneider PC (oh, the good old days :-)). I don’t know why, but the mere phrase “Artificial Intelligence” resonated with me and I made a decision on the spot: I want to take a course on AI. Recent posts and opinions from industry leaders and great thinkers alike - people like Elon Musk and Stephen Hawking - take a very different perspective on AI. When they express their opinions, they usually refer to strong AI, or a computational system capable of outperforming human beings in EVERYTHING. My opinion, based on 23 years of learning and research in the area, is that the future that they envision and that the media often talks about is still decades away. Humans are a truly remarkable computational system, and despite warnings about how AI could end humankind as we know it, I think (unfortunately) that we are likely to be much better at doing that ourselves.

On more immediately realistic grounds, pragmatic AI provides a very useful set of tools and capabilities that can be helpful in our day-to-day lives, including in the world of business. For example, intelligent chatbots are one of the areas that AI is driving forward by providing more human-like interfaces, whilst also reducing the time that people spend delivering business analysis. Imagine the value of an analytics chatbot with the power to answer questions like: “Where are our most valuable customers located?”, with a map that accurately pinpoints these customers’ locations. And then imagine being able to follow up on that conversation with a “by country” overview, instantly getting a table displaying the aggregation of key customers by country. The time saved from avoiding individuals having to manually research, generate, analyse and present such information could be worth a lot to a wide range of businesses.

Another great example of how pragmatic AI has the potential to help businesses is in how it can help them to better understand and serve their customers. Advances in machine learning (a sub area of AI) and big data infrastructures make it possible to analyse and understand customer behaviour and use it in several ways, like helping banks to forecast the credit score of new customers, or typifying telco customers so that marketing campaigns can be more efficiently targeted, or even personalising customer interactions by providing intelligent recommendations specific to that customer. A recent event that demonstrated the growth of machine learning was the LxMLS (Lisbon Machine Learning Summer School), at which more than 300 attendees from all over Europe and North America gathered to learn about some of the most recent advances in the area. Indeed, CRITICAL Software were present showcasing some of our own demos and sharing our expertise.

One of the main challenges for AI, in particular natural language processing, is the ability to quickly extract useful business information from documents. More than 80% of information typically stored inside a business is tied to some form of natural language, whether that be in documents, presentations, emails, notes, chat conversations and other such examples. Businesses have potentially huge, information-rich assets that they can’t use because it would take an overwhelming amount of human effort to process in a useful and systematic way. But, through natural language processing and the computational power that is available today, this no longer has to be the case. Whilst it remains a complex and very domain-dependent task, AI has already been providing huge benefits to businesses that have invested in solutions to these problems.

These are just some of the examples in which pragmatic AI can help businesses to be more efficient. Through avoiding the hype train and applying the power of AI in a focused and practical way, we can take small steps along the long path towards making our future better.

Want to know more about our work in AI & Machine Learning? Visit our dedicated page.