AI may not be new in itself, but businesses across verticals and size, are increasingly adopting AI technologies for several functions. ITPV takes a look at what is driving this trend and what AI applications we are likely to see in the near future.
Most people associate AI with technology giants. However, AI is not exclusive only to those with massive capital. Many small and medium sizes companies are dabbling with AI, developing applications that have immense potential to positively impact people’s lives at various stages. We examine some of these applications in this article.
From Google Assist to self-driving cars, there seems to be a sudden surge in interest in Artificial Intelligence (AI) from both, media, as well as technology companies alike. However, the endeavor to make machines ‘think’, rather, take different decisions based on different inputs, instead of rigidly following a set process, has been going on for decades now.
AI – A work in progress
“It’s only the positioning of AI that has changed, from the high-end sophisticated industrial applications that they were exclusive to, now to routine or mundane end-user applications which touch our lives directly”, says Naveen Kashyap, Managing Director at Yokogawa IA Technologies India Pvt. Ltd. He explains with the example of the (now) primitive T9 dictionary we used in our mobile phones generations ago, or how even the early social networks ‘suggested’ friends a decade ago. It was artificial intelligence at work, albeit at a rudimentary level, compared to what we have now.
Agrees Suman Reddy, Managing Director, Pegasystems India, saying, “Today, the average person knows about AI though things like self-driving cars. AI in our industry (CRM) goes back many years. We have been implementing sentiment analysis on what our customers’ customers are saying in various social media.“
In fact, AI has been used in industrial automation and high-end and mission critical applications for much longer. For instance, the first demonstration of auto-pilot for aircrafts was done more than a hundred years ago using purely mechanical components.
So, why does it feel like an AI explosion now?
The last couple of years has seen some significant milestones crossed for AI in the consumer technology front, no doubt. Naveen Kashyap says, one should also note another reason for the AI buzz today: what’s also changed now is how technology companies are positioning and evangelizing AI, leading to greater coverage by mainstream media (technology publications have been covering AI for well over a decade now). This is in tandem with end-user products / services like Google Assist, which was launched with a lot of fanfare recently. Continuing with this example, its predecessor, Google Now, has been around since mid-2012.
It’s almost as if technology companies have a new found confidence in launching AI based products and not trying to mask the role of AI in them. Until now, AI did not resonate positively with most people, being seen as something that will take away jobs and worse still, lead to robots taking over mankind. Today though, nearly everyone with a smartphone is able to see the benefits of a machine that can think, and AI seems to be quickly shaking off much of that reputation.
Data Explosion and Compute Power
Naveen Kashyap attributes the more recent developments in AI to two other technology related trends – availability of data (enabled by digitization, IoT, etc.) and availability of greater computing power that is needed to process all that data.
For AI to perform any role, there first needs to be massive data related to that role. Data collection has been happening for a couple of decades now since the start of digitization / digital records keeping. Today, there is sufficient data related to many business processes, using which learning models can be successfully constructed. The fact that there is tremendous interest in big data analytics today is an indicator of the massive amounts of raw data that we have collected, and are collecting from multiple sources.
Running complex algorithms to make sense of available data and make decisions in real-time needs immense compute power. Employing the increasingly fast as well as power efficient CPUs, it is becoming feasible to run AI applications even on handheld devices.
In the foreseeable future, the availability of data, as well as compute power is set to grow, thereby enabling AI developers to build more robust, and greater variety of AI applications.
AI for Everybody
Traditionally, AI was seen as going hand-in-hand with high-end cutting-edge technology like industrial automation, robotics, in nuclear power plants and such, and not quite as something meant to solve smaller problems. Today, it’s a very different environment. Even modestly set up technology companies can dabble with AI, so long as they have access to data on the process they want to automate, and the required compute power (discussed earlier). The differentiator is the ability of the developers / data scientists to understand the data relationships and build intelligent and robust algorithms. Not always does this require massive infrastructure investments.
Many of these applications can have tremendous impact on our lives, and these aren’t limited to the likes of driverless cars, or drones delivering packages. There is great potential for AI based applications to bring about socio-economic change at the grassroots. Let’s look at how a relatively simple AI implementation for farmers would work, using AI in the background.
Naveen Kashyup cites an example of an AI powered application for farmers for fertilizer advice as follows: the application would work by identifying a farmer by say, the mobile number, and in real-time lookup his history, when a query is made. For example, to suggest optimum use of fertilizer, the application analyses the crops that the farmer has previously sowed, the type and amount of fertilizer used earlier, the soil type, recent soil test results, prevalent weather cycles, yield data from his farm & other matching regions etc. Based on these parameters, a type and amount of fertilizer is suggested, along with information on the closest locations for procuring it at the best price, along with government subsidy, if any. Of course, the data mentioned above has to be first available, but once the initial hurdle is crossed, the application gets better over time. By analyzing large number of use cases where a certain advice led to better yield under similar conditions, the future advice keeps getting better. Similar applications in precision agriculture, shrimp / aqua culture have been delivering promising results already.
Another example of AI aiding in brining better quality of life is in healthcare. Hospitals in India are already employing AI in oncology care for diagnosis and drug matching to personalize treatments. While this ensures better treatment outcomes, another significant aspect is that it would aid in ensuring that the reach of cancer care is widened.
AI enables hyper-customization
As seen in the farmers’ application hypothesized in the earlier section (see above), one of the most – or, perhaps the most – ground breaking feature that AI provides at an individual’s level is that of extreme, or hyper-personalization. Nobody loves the ability to create highly personalized profiles of individuals as much as sales and marketing people do. No wonder then that one of the biggest use cases for AI in businesses currently and in the near future is in enabling marketers to create highly personalized campaigns.
E-commerce companies are already using AI to ‘suggest’ next purchases to a customer based on previous browsing and purchases. This application relies on mostly structured data. What if AI could work on unstructured data, find ‘context’ to data from different sources, and generate a profile of a ‘person’, more than just a pattern of buying habits? The use cases for such a tool are aplenty in businesses.
Today AI powered tools are in the works which can look up an individual’s presence on social media, news networks, etc. and present a summary of that person instantly, thus enabling salespersons with usable information on their prospect, before they head to a meeting. Rather than walk in blind to a discussion, they are able to break ice easily, and quickly converse on the matter of interest.
Marketing teams can quickly select the most apt invitees for an event, from a large pool of individuals in a database. Instead of scrolling through huge lists manually, which is an error prone process, they can just enter in desired profile, (or, area of interest of the invitee) in the application. Even if such data is not manually fed into the system, it can figure out based on the individual’s mention in news and participation in social media.
This isn’t the biggest impact coming from hyper-personalization. That claim would go to customer support and allied services. Apart from the obvious business driver of keeping cost low for customer service, since it is purely a cost center, there is a second, bigger reason for deploying AI commonly in CRM applications. It is to do with the massive change in expectation from the millennial generation. There is an expectation of super-quick service, as well as personalization. Nobody wants to repeat their case history to different customer care executives every time they make a call. Or even, having to identify themselves and the service they consume, for that matter.
What if, the CRM application could predict what the customer is calling for? Or, suggest a highly relevant upsell based on the customer’s profile, instead of a blanket spam-like promotion that happens today? According to Suman Reddy, CRM applications already are able to suggest a ‘next best course of action’ to customer service representatives, based on the caller’s (customer) profile from previous interactions, as well as similar cases in the past.
Taking AI in CRM a step further, chat bots can effectively replace a human chat agent for routine tasks such as address update, change in service plan, etc., says Suman. These use cases are being actively explored right now, with many small and large technology companies developing AI based CRM tools.
Business leaders embrace AI
Infosys recently shared a report titled Amplifying Human Potential: Towards Purposeful Artificial Intelligence, by polling 1600 business decision makers across the globe on how they perceive AI. The findings leave little doubt on the increasing relevance and adoption of AI for various business functions.
- The most telling figure is that 76% of respondents agree that AI is fundamental to the success of their organization’s strategy. On a related note, 71 percent believe that AI is “inevitable”.
- ‘Gaining competitive advantage’ was the single biggest reason cited by respondents who said their organization uses AI in some form (28%).
- 25 % of respondents who use AI, said the move to deploy AI based tools was led from an executive level.
- Further, 25% of respondents already have ‘fully deployed” AI technologies which are working as expected. In the same breath, 81 percent said they have some form of an AI technology implemented (but with varying results).
- The majority (69%) report that IT is using AI. The top three departments that are open to using AI are IT systems and security (54%), data analytics (43%) and customer service (43%).
Challenges businesses see for adopting AI technologies:
- Fear of change amongst employees topped at 54%, with another employee-related issue of (lack of) cultural acceptance at 47%.
- Lack of in-house skills to implement and manage – 54%
- Surprisingly, almost half of respondents (49%) cited lack of knowledge about where AI can assist as a barrier.
The learning from these numbers are clear. In an increasingly competitive world, gaining competitive advantage is a top priority for businesses in most verticals, and this being seen as the single biggest driver for AI adoption leaves no doubt that AI is set to make inroads into businesses.
Change management and education is a crucial part of any AI adoption project, since fear of change, cultural acceptance, and even fear of handing over control were each reported by about half of respondents as challenges.
Opportunities for channel partners
Channel partners can bring value to their customers by playing a consultative role, as well as executing / deploying AI based solutions for them, either as standalones or as part of larger projects. Those channel partners who already play such consultative roles and therefore understand their customers’ business processes are at an obvious advantage here. Once inefficiencies or scope for improvement is identified, it is then a matter of carefully identifying suitable AI technologies. Since AI is not about just deployment, and is a continuous improvement concept, choosing the right vendor is critical, who needs to be involved with you even after deployment.
Before taking on AI project, it is prudent to explicitly explain to the client, the importance of data purity, and the fact that AI, by its very nature will get better over time, as more and more data is collected and analyzed. AI cannot always guarantee ROI from day one.
by Kailas Shastry