Your browser is out-of-date!

Update your browser to view this website correctly. Update my browser now


AI: driving development

Picture: Yvonne Albinowski

In the first part of this feature we outlined the enormous potential of AI in AV while also highlighting the security threat. Here Steve Montgomery considers the areas and applications set to benefit from future developments in AI.

Recent trends for customer support, location and demographic based advertising, and geo-referenced imagery have driven the development of the technology. These will lead us into newer application areas in which the AV community can derive benefit from artificial intelligence technology. Yiannis Cabolis, director, technical innovation, Electrosonic, believes that cognitive applications, advertising and branding, point of sale (POS) systems, health and hospitality applications will experience increasing benefit in the future. Already large organisations, including IBM, MS Azure and Google are investing heavily in creating tools that can access broadcast feed metadata to detect important information and produce breaking news feeds and that can automatically identify relevant sources, hot topics and event-specific content for use in signage systems.

Innovations around AI have turned the chatbot from an annoyance to an integral part of the user interaction experience. Even the simplest bots, so-called ‘rule-based-chatbots’, have come on leaps and bounds. Rule-based chatbots are now able to hold basic conversations using ‘if/then’ logic. “There are more than 300,000 Facebook chatbots in action,” says Ahmed Helmy, CTO, Avaya International. “They are mostly used to automate customer service, online sales and marketing.  They communicate with users using a call to action button and they’re proving extremely effective at resolving simple queries. Many routine requests for information can easily be handled by a rule-based chatbot.

“Things are moving beyond rule-based bots very quickly with the adoption of ‘real’ artificial intelligence. This is what you see with what we call natural language processing (NLP) bots. NLP assists machines understand human language, so instead of the visitor having to navigate through buttons and menus, they can simply have a conversation with the bot in the way they would message a friend. What’s making them even better, though, is the addition of machine-learning (ML) into the mix. This makes bots optimised for learning about the visitor or user, retaining information on them, and predicting the next steps in a conversation in much the same way that computers pre-fetch data based on previous interactions.”

“With the proliferation of AI tools there will be a whole new vertical that will fall into the IT/AV scope of work to support and resolve security concerns”

This evolution will take us to the point at which we will be able to converse with bots naturally and use them to perform relatively advanced functions. Control and interaction of AV systems will then become a matter of simply talking to a system in order to get it to do what we want or provide information to us.

Artificial intelligence is likely to present us with a whole new security threat in the coming years. Cabolis adds: “With the proliferation of AI tools there will be a whole new vertical that will fall into the IT/AV scope of work to support and resolve security concerns. However the end-user will also have to be aware and educated on how to live in a world rich of AI. We all are aware of our digital/IT footprint but things can become a lot more complex. AI’s propensity to misrepresent and manipulate data, interpreting and miss-interpreting individual decisions and eliminating human interactions and decision making or control are very real challenges to its capability.

“Planning to overcome security concerns will require software to prevent cyber attacks and secure IT infrastructure and data acquisition technologies. Just as major department stores, financial institutions, and social networks have recently had major security breaches which lead to the exposure of personalised data on the web, artificial intelligence will be no different.”

Gaining recognition
Facial and voice recognition certainly make it easier for companies to do business. Apple has proven that with its Face ID technology, and the leading consumer-focused companies are using AI-powered voice biometrics for customer authentication in the call centre.

There are natural concerns around security and privacy of facial and voice recognition and companies should certainly be aware of them. According to recent research, 80% of global users believe that organisations are not handling their data securely. But by the same token, they don’t want privacy policies to impact convenience: 74% say they’ll buy more from organisations that make it easier to do business. Users overwhelmingly want convenience; most say this is more important than price. So while security and privacy should be a high priority, it should not be a barrier to the adoption of AI-based technologies.

“Face-detection systems that identify a type of person according to age, sex and ethnicity are particularly useful in the AV industry, particularly in digital signage applications.  They enable audience analysis to be undertaken and tailoring of content to match the viewers at that time,” explains Alan Hopkins, director at Visionpoint. “This data is captured anonymously with no personal information or images held or stored and therefore is not subject to current GDPR rulings.  

However, the differing technology of facial recognition which matches an image captured by a camera to an actual person using a previously recorded profile held on a database is fraught with issues around data protection and privacy. The image has to be captured, encoded and then matched against stored images. That requires highly-secure encryption so that images can be transmitted and stored across systems, all covered by GDPR regulations making it a very challenging activity. In addition significant image processing is necessary for these systems and with the current level of computer technology this usually requires fast, expensive graphics cards (GPUs), similar to those used in high end gaming PCs, which makes it difficult to deploy at an economically-viable level.”

AI assistants are already commonplace and will become more so. In some cases such as in cars, totally hands-free smartphone interfaces are likely to be part of future legislation. Already, voice entertainment interfaces abound at home, in the car and there are some examples at work. Alexa and Siri are driving the growth and acceptance of these types of devices throughout our lives. “Virtual assistants are already common and they’re only going to become more pervasive as their functionalities increase,” points out Cabolis. “Apple’s Siri and Amazon’s Alexa make for great examples.  Both are essentially NLP bots. Their AI capabilities include the ability to understand human language in the form of voice and text and give intelligent replies or carry out certain tasks.”

We’ll continue to see organisations hone and refine the capabilities of the AI systems already in place today. In the future data will be collected from a much wider range of different domains and industries, hopefully collected and exchanged on highly regulated platforms. This will securely link enterprises and significantly expand the efficacy and expertise available, thereby enabling organisations to deliver an exceptional level of service