Artificial Intelligence (AI) technologies will become widely available and commoditized
By Rajesh Kumar R
The technology eco-system is rapidly evolving to enable AI technologies become more developer-friendly so that enterprises and the tech community can adopt them faster. More importantly, technologies that were earlier confined to scientists, research communities and universities, are now easily accessible to any engineer who can code at an affordable price, allowing widespread usage and production.
Technology giants such as Microsoft, Amazon and Google are in the AI race as never before which is expediting and attracting huge investments in technology that is widely accessible ‘as services’ to the rest of the world.
Another important aspect of this development is that these tech giants are also cloud infrastructure providers and hence the need for GPUs, and large computing capabilities to work on these technologies are now becoming irrelevant. This infrastructure is available as a service on a server-free environment thereby condensing the need for it. In addition to the above factors, commoditization and proliferation of these technologies is being aided by lower price points they will be offered at.
Some examples of AI technologies that will be available are vision technologies such as people recognition, object detection, Natural Language Processing (NLP) technologies, and voice technologies such as speech recognition and text to human-like voice.
Applications of Machine Learning (ML) will become ubiquitous
Machine Learning (ML) is one of the important components of AI technologies which will see extensive use. When Big Data technologies became popular few years back they were largely limited to number crunching and synthetization of large volume of data with statistical models that led to prediction models, forecasting and so on. However, with the trend of commoditization of AI technologies, ML brings in a brand new way to deal with the enormous amount of data which is out there and is also an immensely powerful way to put the data into use.
With the earlier model, data scientists and big data technologists (who were often extremely expensive to hire) were the ones who could deal with a large amounts of data and make sense of it. Now with ML, this data is fed to the machines which learn, glean insights, and create algorithms and models.
As long as large volumes of data is available to learn from, applications will be able to leverage and use ML to make the solutions smarter and data driven. For example, simple help desk applications will start to route the problem tickets automatically based on the historic data to the teams such that they get fixed at the earliest; grocery stores will get demand forecast when an adverse weather system is developing in the area; hotels prices will be determined based on an event in a city even before bookings open; airlines will know when a flight is about to get cancelled so that they can reschedule.
Robotic Process Automation (RPA) will become a norm for executing business processes and they will get lot more smarter – powered by AI technologies
Currently, there are numerous RPA implementations in practice for automating business processes. Many organizations providing BPO services are driving this automation wave by themselves. While more investments will be made to achieve the desired level of automated outcome, usage of RPA will eventually be taken for granted while establishing the business processes, which will largely impact the way the work force is currently deployed.
Essentially RPA will move from being an initiative to being a ‘way of life’ for most enterprises. The work force will quickly change from being business process execution to a team of Subject Matter Experts (SMEs), who will engineer various processes. Automation engineers will implement these processes with the help of bots and enterprises will have a small set of knowledge workers for processes where human judgement and decision making is essential.
While RPA will become a norm, they will also become smarter with prevalent access to AI technologies, where past data will drive the automated decisions leveraging ML capabilities and technologies such as Optical Character Recognition (OCR) to read scanned documents, voice technologies to answer phones and provide variety of services. The business process execution will see massive amounts of automation powered by RPA and AI in the coming year.
User Interface will evolve into ‘Natural Language’ Interface with chat and voice
While the power of computers have been increasing exponentially, the user interface is also evolving along with it, albeit at a much slower rate. In the earliest computers, the man-machine interface was primary driven by the machine rather than the man where interfaces such as punched cards were used to interact and program. This eventually evolved into a text based interface and with the PC revolution, graphical user interfaces (GUI) have become popular. With the advent of web based applications the experience of GUI is becoming better and richer. However, the primary mode of interaction remained to be the key board and mouse – which one can argue is still machine driven rather than human driven. Nonetheless, the evolution of GUI itself took a huge leap when touch based interfaces were introduced with the smart phone revolution, which can be claimed as intuitive and human friendly. This largely eliminated the need of user manuals and training to use an application with the touch interface taking the centre stage.
The next big leap in user interfaces will be chat-based UI which will further evolve into voice based interfaces. Essentially, applications and solutions will be provided with no user interface for the application itself will be made available over multiple channels via chat based solutions. For example, a leave application is typically logged into the intranet, navigated to a link and eventually picks the date, leave category. The authorizing authority follows the similar procedure to reject or accept the request as well. With chat based UI the process is simplified – the employee can connect with an instant messaging system and chat with a bot, saying “want to take a personal day off tomorrow”; the bot will immediately intimate the manager via chat message which can be approved using the same interface. Thus, the entire process can be closed within a matter of few seconds.
In addition, this kind of interface provides a user friendly channel that can abstract various applications and systems over a single channel. Additional features can be added seamlessly without the end user having to install or update an app or for the system admins to rollout a software upgrade to the desktops.
In conjunction with ‘Natural Language Processing’ (NLP) technology that is powering the chat based solutions, another adjacent technology on voice recognition and text-to-speech is also evolving swiftly, which will convert chat based systems into voice enabled system – case in point would be ‘Alexa’. These voice based systems currently at the consumer end will soon hit the enterprises. Under these circumstances, most of the software applications in the enterprise would be rendered over the widely used instant messenger and bots could be answering the support calls. The current frustrations with the IVR systems will vanish as bots will sound and understand more like humans. So after all, the urge to press 9 in an IVR hoping to get to a live agent will no longer be a necessity.
Cameras will become smarter and will recognise people and objects, leading towards proliferation of video based real-time solutions, will become part of the larger IoT eco-system
With the computer vision technologies now becoming commonplace, cameras will increasingly start to have this computer vision technology bundled into them, enabling the advent of smart cameras which detect and recognize people and objects. Today, computing power is available in small packets which allows devices to become self-sufficient and work autonomously without having to connect to the cloud in order to access these recognition capabilities. So, in the near future video based analytics, event or data driven surveillance systems and many more applications that are based on real-time video feed will evolve.
These capabilities of cameras recognizing objects and people combined with their ability to access the network cloud leads us to extensive application of Internet of Things (IoT) technology. IoT solutions will now not only have the power of smart edge devices but will also leverage the AI ecosystem available over the cloud forging technologies and solutions that are unprecedented.
The author is the Head of Automation and Global Delivery for Retail, CPG and Manufacturing, Mindtree
