(part 1 is here)
The structured, processed and cleansed datasets become the notorious Big Data after algorithmic analytics applied. In so doing, the volume of the initial “raw” data can be much less, than it is used to be required from the huge data libraries. The well-governed 2Gb can be much more valuable than 10 Tb of the semi-structured and irrelevant data especially in terms of a particular business task the data was collected for. But who can figure out which dataset is relevant and valuable for a certain technological or business task? And how much data is in fact necessary to fulfil the business agenda of an organization?
Answering those questions is not so straightforward as it may seem at first sight. In many cases, the organizations slip into the naive romance while starting to use the Big Data practice. Either deliberately or due to the lack of experience, they overload data scientists with the responsibility regarding the outcomes and business decisions to be made about the received data analytics.
The big bosses expect the newly hired data analysts using the expensive and so advanced deep learning technology should perform a miracle adding the extra value to the business that undoubtedly will cover the investments applied for the deep learning hardware-software complex along with data libraries. At worst, the organizations are trying to manage without data specialists at all assuming that the very availability of Big Data is enough to push their business forward.
The cart before the horse
Some may argue that the business romanticism is inherent only in the young inept organizations and startups having no clear idea about the deep learning technology. However, the examples of confusion about what to do next with the available AI and deep learning include even the business monsters. IBM Corporation put about $100 million investing in the app development for their Watson cognitive computing system in 2014.
After winning Jeopardy in 2008, Watson looked like a state-of-the-art go-to guy for achieving new business goals. Those 8(!) apps available now on their homepage can hardly compose the valuable outcome of the investments.
It seems that Facebook and Google releasing their advanced algorithms and AI engines (Tensor Flow) for free are going to attract some fresh brains and ideas from outside while their native talents are unable to transform the technology into an income sufficient for covering the investment at least. Probably, the mere availability of AI and deep learning is ahead of their business appliance capable of adding extra values.
May be the initial absence of a clear understanding that the algorithms and Big Data as such could not be the ultimate goal of business has brought the Silicon Valley guys to that free-access solutions finally.
And apparently, the undertone of Sergei Brin’s statement that he does not see how deep learning comes should show that the further evolution and deep understanding of AI business applications are still expected.
Business pragmatism vs cargo cult
In order not to transform Big Data, algorithms and deep learning onto a cargo cult the businesspersons should realize two following pragmatic ideas:
- Any collectible data processed by the deep learning algorithms is valuable for a business only in case some definite changes happen as a result of the data analytics implementation. Moreover, the changes should provide some positive agenda adding extra value to the business. If the data analytics does not lead to any business transformations that can increase incomes or optimize the workflow, the whole process including data libraries, deep learning algorithms, hiring data scientists and so far is simply meaningless. Although this simple idea looks obvious, many organizations overlook it being enchanted by the apparent omniscience and power of the aggressively promoted AI and deep learning solutions.
- The data collection, ingestion, and manipulation, as well as modelling and visualization, belong to rather statistics as to the process residing between the business tasks and the actionable insights, valuable products or services achieved from the scope of activities implicated in transforming collected data. It means the quality of data depends on the clear understanding what feasible actions are required following the current business task. The origin of the received data analytics does not matter. The datasets and software can be either bought off the shelf or developed by the own staff. It is counterproductive to expect a miracle from any deep learning while a business does not understand to which final objective the data should lead. And despite any resistance of the AI adepts’ camp, the only intelligence capable of keeping up with the organization’s goal-setting system is namely the human intelligence.
A piece of cognitive dissonance
In the light of the recent impressive developments in the artificial intelligence field such as the 96%-accuracy image recognition or winning Go game, many forget about the main goal of the AI-powered solutions implemented by the Internet-based companies. The cornucopia of profits on which Facebook and Google (for instance) are grounded is getting the Internet surfers to view and click the ads.
However, being extremely efficient in figuring out which content and media are to be prioritized for each definite Internet user, the AI-powered assistants (such as Siri or upcoming Viv) come into conflict with the above-mentioned goal. Suggesting content consumers about the media they like most of all, the AI-powered assistants allow people seeing much fewer ads. In doing so, the owners of AI solutions disrupt the entire advertising industry. The very industry that provides them with income! Are they really getting themselves hurt?
Such a mismatch looks like a mess rooted in a wrong goal-setting system inherent in the contemporary craze among AI adherents. It seems the digital giants go too far betting on their AI. Indeed, a lack of the human intelligence is clearly observed.
Turning AI into IA
The artificial intelligence movement is exposed to many futuristic forecasts and predictions. While professionals tend to discuss the next technological Leap based on the overall digitization in order to implement AI into each and every aspect of human life, the vast variety of dilettantish observers prefers speculating about different sinister scenarios of possible enslavement of the human race by AI (regards to Matrix movie). Nonetheless, both camps start from the joint premise that the AI-powered solutions can evaluate to the degree sufficient for shifting people away from the majority of jobs and activities that humans are involved in now. The investment amount along with the extent of development efforts attempted by such business leaders as Uber, Google, Apple, and Amazon hint at the actual social problem of the enormous unemployment people can face rather shortly.
In order not to confront with the seemingly inevitable robotization, some social activists propose such solutions as the Universal Basic Income as a method of alteration of the entire wealth distribution system. Thus, the problem is moved from the technological to a social paradigm. However, there is a hybrid approach combining the technological alternative and the social responsibility of the AI developers.
The approach implies that instead of replacing humans from some activity the similar machine learning technologies as used for AI can be applied to assist people in more effective fulfilling the objectives. The so-called intelligent augmentation or IA aims to get the best of both worlds. Combining human brain and machine learning, IA can eliminate the mess of the business goal-setting system described above. Besides, the organic fusion of the artificial and human intelligence can result in fostering of the social security making people less vulnerable to unemployment.
From the technological standpoint, IA is hardly more complicated and problematic for developers than AI. Indeema hopes the strategic anticipation regarding the feasibility of intelligent augmentation will prevent the global digital giants from underestimation of the business profits and social privileges that could be brought to humanity by IA.