AI has its place in business, just don’t believe the hype
Business Management

AI has its place in business, just don’t believe the hype

Artificial Intelligence (AI) is getting a lot of attention these days, but like the Internet of Things (IoT) and cloud computing before it, a great deal of hype and marketing is already starting to obscure just what AI is and what the potential benefits could be for businesses.

AI is by no means a new sector of technology, but it has spent long years in the wilderness after repeated failures to deliver much on its promises. Now, new approaches are beginning to bear fruit, and AI seems to have become another buzzword that is thrown around with abandon by technology companies trying to add some extra glamour to their products.

At the giant CES show in Las Vegas last month, virtually every product on show had to have some AI spin to it. Thus we were treated to an AI toothbrush intended to improve your dental hygiene, an AI hairbrush to monitor the health of the user’s hair, and that evergreen trope of the consumer technology industry, the connected fridge, was also given the AI treatment.

As innovative as some of these devices may be, their operation more often involves conventional pre-programmed algorithms than anything that could genuinely be described as artificial intelligence. And many systems that are described as using AI, such as recommendation engines in consumer retail applications, often appear to do little than simply look at the user’s previous purchases and recommend more of the same.

But all of this hype conceals some real advances in the way that AI can now be used in order to improve the accuracy of predictions and forecasts, or produce better results without being explicitly programmed how to do a task, and IT companies are looking at how these approaches can be blended into business applications and processes.

Microsoft, for example, is enthusiastic about AI, and chief executive Satya Nadella spoke last year of “democratising access” to AI to help solve challenges, which would involve “infusing AI into everything we deliver across our computing platforms and experiences.”

This approach can be seen in a recent blog post in which Microsoft showcased how deep learning techniques can be embedded inside its SQL Server platform. This uses the company’s implementation of the R language for statistical analysis to produce results, and in this specific example, consists of automatically recognising and classifying images into distinct categories.

However, real-life applications of AI techniques may include risk analysis and fraud detection in financial services firms; predictive maintenance of equipment in the industrial sector; and predictive planning of inventory for retail outlets.

But what do we mean by AI? In its simplest form, modern AI means creating systems that are capable of learning from examples, and then making predictions when presented with new data. The key feature is that the system is capable of making predictions that it has not been explicitly programmed to come up with.

Those with long memories may recall knowledge-based systems, an earlier AI approach that attempted to solve complex problems through the application of programmed rules to a knowledge database of facts relating to the particular problem.

In contrast, modern AI depends more on statistical algorithms and data – sometimes lots of it - to train the system to produce the desired result from the available input. This is called machine learning, and is behind many successful AI platforms, such as IBM’s Watson.


Chatbots as first line of customer contact

Although the name is associated with a computer system that defeated human contestants in the TV quiz show Jeopardy, the technology behind them can be used for more practical purposes, such as fraud detection or aiding medical diagnoses.

“There tends to be a bit of a misunderstanding that there’s some kind of big supercomputer called Watson and it does some kind of magic,” said Duncan Anderson, European Chief Technology Officer for IBM Watson.

“In reality, it’s a set of discrete programmable APIs that do different things, so there’s one that does image recognition, there’s one that does understanding the intent of a question, and there’s about twenty of these that we have. When you build a solution, you may use two or three of these and stitch them together to solve a problem. It’s much less about big supercomputers and more a collection of different capabilities that you bring together,” Anderson explained.

One area where Watson is finding interest from businesses is for chatbots, automated systems that are crafted to handle customer queries in natural language. IBM announced a pilot with Royal Bank of Scotland in October for such a system, which will be able to respond to simple requests or handover to a human operative if the query is too complex.

“You can chat with it and ask it questions like ‘how do I apply for a mortgage’, or ‘I’ve lost my credit card, please help me’, and because it is dealing with a very narrow domain of knowledge, it can be very good at it,” Anderson said.

“We think that lots of enterprises will be adopting these kinds of narrow domain-specific virtual assistants for a number of reasons. Firstly, the cost of these things is quite low, and it could also generate more customer interaction, because if you can chat via messaging with an organisation this way, you might ask things you wouldn’t otherwise bother with because it would be too much hassle to get through to someone on the phone,” he added.

Another area where the technology is being used in the finance industry is for fraud detection, by examining transactions for patterns that may indicate suspicious activity.

“Lots of financial institutions have various markers intended to flag up suspicious activity, such as an account being used to purchase a couple of iTunes tracks just before a big purchase. This is typical of the way a criminal may act; you do a few very low value transactions to test if a stolen card is working, and then you buy something expensive,” Anderson explained.

“That concept is beginning to expand out to other areas, as people realise that if you can predict fraud, then what else can you predict? Can you predict spending patterns, for example, where someone may be in a distressed financial state because they have certain patterns on their behaviour that indicates that it should be looked into?”

Meanwhile, the same kind of pattern recognition capability is being used to boost cyber security, with IBM announcing a beta programme in December that uses Watson technology to identify suspicious behaviour, but also to better determine whether or not a current security incident can be attributed to a known malware tool or associated with a particular cybercrime campaign.


AI will help, rather than replace, humans

The next level on from machine learning is deep learning, which makes use of large-scale artificial neural networks to accomplish its pattern recognition tricks.

“Machine learning generally speaking means an algorithm that works at a relatively superficial level, whereas with deep learning, you typically have a neural network that has quite a few levels within it,” said Anderson.

It is these complex layers of neural networks, combined with the use of large volumes of data passed through the system to train it, which can deliver some quite remarkable outcomes.

“Deep learning is essentially something that we could easily confuse with human intelligence. If you need something that appears to be intelligent, that is very nuanced in its behaviour and can identify very complex situations, that is probably a deep learning system,” Anderson explained.

These capabilities are now driving natural language translation tools such as Google Translate, which is getting remarkably good at taking words, phrases or sentences in one language and automatically translating them into another.

Another example of the technology is image captioning, whereby the system not only has to identify elements within a digital image, but also infer what is happening, so that it can come up with a natural language description such as ‘a man walking down the street’.

In fact, computer systems based on AI techniques are now becoming so adept at certain tasks that there has been a growing concern that more and more jobs for human workers could be lost to automation. In particular, it has been suggested that more professional roles that deal with complex information, such as lawyers, could be at risk of being automated away.

IBM is dismissive of this view, and believes instead that AI will help workers to do their jobs better, by automating the boring tasks and highlighting areas that need human attention. Indeed, IBM chief Ginni Rometty made this clear during the World Economic Forum in Davos this month.

“We see a partnership between the human and the artificial intelligence. The AI is supportive of the human, it’s not trying to take over the human’s job,” said Anderson, echoing his chief executive.

“I think the way we see things is that AI brings huge opportunities, and different skill sets will be required, the nature of work will change, but new roles will emerge,” he said, “including new roles for workers to configure and train AI systems.

“Some of the simpler things will be automated, but humans will end up doing the more complex situations, and we see it as a case of evolving rather than these more apocalyptic scenarios, very much an opportunity for society to solve some of its more complex problems.”


Also read:
What an unrestrained AI might bring


«What is the impact of Africa’s (many) internet shutdowns?


Latin America startups aim to corral the world’s cattle »
Dan Robinson

Dan Robinson has over 20 years of experience as an IT journalist, covering everything from smartphones to IBM mainframes and supercomputers as well as the Windows PC industry. Based in the UK, Dan has a background in electronics and a BSc Hons in Information Technology.

Our Case Studies

IDG Connect delivers full creative solutions to meet all your demand generatlon needs. These cover the full scope of options, from customized content and lead delivery through to fully integrated campaigns.


Our Marketing Research

Our in-house analyst and editorial team create a range of insights for the global marketing community. These look at IT buying preferences, the latest soclal media trends and other zeitgeist topics.



Should the government regulate Artificial Intelligence?