Cogito—Arago sum?

Successful companies gather, analyse and store data, but still find it challenging to share this within the organization. Nevertheless, the focus of machine learning should be primarily on delivering and exchanging the right knowledge in an effective manner.

In a recent paper, technology research firm Gartner Inc. suggested that we begin using the term ‘smart machines’ instead of the plethora of terms now being bandied about to describe new technology. Its logic is that the term ‘smart’ is less objectionable than other words such as artificial intelligence (AI) and cognitive computing. Machines, it says, cannot think or reason, and ascribing such terms to them gives them an anthropomorphic quality, which in simple English means that we start thinking about these machines as equivalent to human beings. 

The research firm claims that this anthropomorphism gives rise to what poets call the ‘pathetic fallacy’, which assigns the capability for pathos, or deep feelings, to machines that are devoid of any capacity to feel. The paper seeks to clarify many of the terms associated with today’s computing technologies, and warns information technology (IT) leaders against being seduced by such anthropomorphic terms. It suggests the term smart machine, which while not perfect by itself, at least begins to dial down the hype by more narrowly defining the boundaries of new computing technology. The research firm sagely recommends that IT executives ensure that their teams are familiarized with the capabilities and weaknesses of smart machine technologies so that they may exploit the opportunities and ignore the myths.

For the sake of clarity, Gartner defines smart machines as the ones running technologies that adapt their behaviour based on experience, and are not totally dependent on instructions from people since their data crunching allows them to “learn” on their own. This ‘self-learning’ allows smart machines to come up with unanticipated results. However, it acknowledges that these results are sometimes surprisingly good and at other times unacceptable. Ergo, extensive testing is required to get the technology to perform within acceptable boundaries. As an aside, this testing effort could represent a large opportunity for Indian IT services firms that have extremely robust testing businesses.

Smart machines can deliver great results, but they are still machines and are only smart in a narrow sense. They, thus, depend on a large base of information that is essential to their ‘self-learning’ capabilities. To reach this goal, it is necessary to constantly teach machines the essential knowledge, or semantics, that they need to ‘understand’ the purpose and environment they are acting in. This is not dissimilar to preparing children to understand the world they are living in since machines need to be prepared on an ongoing basis. See, despite the warning, there I go: anthropomorphizing again by comparing smart machines to children!

Coming back to machines, the German AI machine learning company Arago says that this kind of training or learning can be achieved with a 1:1 approach or by connecting the machines with other machines or environments. As I have covered before, this may be through the use of 1:1 training in the semantics of common sense, or in the semantics of specialized bodies of knowledge such as medicine, banking, and engineering. 

Chris Boos, Arago’s chief executive, says the success factors of many large-scale enterprises can often be found in their corporate philosophy and work ethic. These frequently contain concepts such as organizational structures, diverse communication channels, and flexibility through responses that can constantly adapt. However, every decision and move assumes a tacit knowledge of the entire organization. When shared and applied appropriately, this factor represents an indispensable and significant company asset that contributes more than 60% of the company’s value, according to Boos. 

Successful companies gather, analyse and store data, but still find it challenging to share this within the organization. Nevertheless, says Boos, the focus is primarily on delivering and exchanging the right knowledge in an effective manner. To start the AI journey, Arago advises the creation of a semantic data pool for the organization to describe the company itself, and the technology and tools used in the company that can potentially be controlled by AI to run many processes of the business. In doing so, Arago recommends starting with IT automation as the strategic starting point since it provides immediate value to the organization, and automating business processes later. The opposite approach, focusing on business process first, is what robotic process automation firms like Blue Prism are trying to do. Arago’s stated rationale to start with the automation of IT instead of business processes is that IT is the heart of the organization where all the data and information and therefore, knowledge, pass through. In addition, a firm’s IT systems themselves continuously generate more semantic data.

The magic sauce, according to Arago, then, would be to collect all semantics, both common sense and specialty oriented, that are germane to the enterprise, and then find a way to disseminate these semantics across the IT pods in the enterprise that support either human- or machine-based business processes.

Arago believes that anything that is a process will be done, or at least run, by AI within a relatively short time frame. Ergo, smart machines have the power to transform any company into an AI-enabled enterprise. However, says Boos, AI-enablement is not a goal in itself, it is just a tool for companies to continue to remain competitive as today’s market morphs into the new age. His company has developed a product called HIRO, which it claims is general problem-solving AI. HIRO, it says, can automate any IT process and later, any business process. IT automation is only the starting point, considering that anything that is a process can be run by AI. 

As a side note, in my experience, enterprises purchase a large amount of different solutions over time, thus building technology silos that are difficult to integrate and highly complex, and so find it difficult to support the business’s goals. It is time IT decision makers considered an AI-related industry-standard approach, possibly the one suggested by Gartner, or pushed by Arago, which starts first within the IT environment and then eventually supports the company’s entire business stack. 

Siddharth Pai is a technology consultant who has led over $20 billion in complex, first-of-a-kind outsourcing transactions. He now works as an advisor to Boards, CEOs, and investors to help them strengthen and execute their global technology strategies in an increasingly uncertain and volatile world.

*This article first appeared in print in the Mint and online at www.livemint.com

For this and more, see: