AI: Does Your Business Really Need It?
Nor is there a precise criterion of when a computer becomes "intelligent’"
From Wikipedia, the free encyclopedia
Whenever terms like "intelligent systems" or "artificial intelligence" are mentioned, mass consciousness brings forth an array of associations.
These mainly come from Hollywood junk movies.
Some real technological things will also be remembered, like — the Turing test, winnings of IBM Watson at chess, or the Google algorithm at the Chinese game of Go.
However, let's have a look at it from the position of a real business and ask yourself a question.
A very boring and down-to-earth one: how can the achievements like the deception of five committee members of doubtful competence, or playing with the traditional Chinese stones help your particular enterprise to earn more money?
It is obvious that an abstract intelligent system will bring a business an additional profit if (and only if) it significantly reduces the amount of human resources consumed and at least does not worsen other key performance indicators.
The additional profit is generated evidently by reducing the cost of labour.
How can the victory in a game of chess over, let's say, Kasparov, lighten the burden of expenses of your enterprise to run your headquarters?
If you ask yourself what that has to do with the price of tea in China, then you are not alone.
In Hollywood blockbusters, AI behaves very human-like.
It talks and displays emotions, has an individual character, wishes and dreams, and can even cheat.
However, any real office (especially when the "bosses" are out) swarms with strikingly human-like biorobots who convincingly demonstrate all the properties of "strong" AI.
The daily achievements of their 100 per cent natural intellect (which are neutrally termed "human factors" in business practice and accident investigations) are something that business owners and top managers would be happy to discard.
Consciousness is not what business system operators are willing to pay for. They would rather pay to get rid of it.
Thus, the objective reality is in complete contradiction to the picture generated through the media by the beneficiaries of "artificial intelligence".
The real business just does not need "strong AI".
"Strong" AI is similar to the human one.
Few people tend to notice that this resemblance involves the copying of shortcomings.
Strong AI, even if it were possible (spoiler: in fact, not), besides the coveted ability to create, would inherit all the weaknesses of its human "parent".
Then why inventing artificial intelligence if the number of those who have natural one has been growing rapidly for a million years every single day without any additional "investment" in a completely natural way and costs the biosphere, nota bene, 2000 calories/ day/person on average?
That is only 2.3 kilowatts per day! Not even the most powerful incandescent lamp.
IBM Summit supercomputer, to understand the scale of the disaster, consumes 15 megawatts.
And the creation of "strong AI" is still not getting any closer.
A real business needs something very opposite — a system with human functionality but lacking the unavoidable drawbacks such as emotion, fatigue, burnout, illness, forgetfulness and corruption.
A system not second to man in terms of functionality but dramatically superior in terms of reliability, discipline, and service availability.
This is our definition of intelligent systems for business.
Let’s cite the example of goods procurement for a retail chain’s distribution warehouse, as one of the simplest, shortest and most illustrative business processes.
At the upper level, it looks as follows:
an employee in charge (category manager) decides that Goods Item A must be purchased (Needs Formulation stage);
(the same or) another employee chooses a supplier and agrees upon the price, quantity, and date of supply (Order Placement stage);
the goods ordered arrive at the distribution warehouse, in the acceptance area, where they are accepted with their quality/quantity checked (Acceptance stage);
the processed goods are distributed to their storage locations (Pigeon-Holing stage).
While traditional ERP systems usually process each stage in an independent "module" (business communication among the employees bypasses the ERP system or is absent altogether) and different "modules"" results do not always match, in the abstraction of IEM systems each employee is at his own stage of the virtual conveyor — that very value creation chain that constitutes the holistic business process.
The category manager creates in the system a "procurement requirement" document that, after his work has been completed, passes into the purchasing manager’s responsibility area (the conveyor stage "order placement").
At his own stage of the conveyor, the manager using the tools of the system turns "procurement requirement" document into "purchase order" with known exact quantities, prices, supplier(s) and shipment dates.
On completing, the document is transmitted to the warehouse acceptance personnel and turns into "goods on order" — a prototype of the future GRN (goods received or receipt note).
After the goods arrive at the warehouse, the receiver checks the accordance of the products received comparing the supplier’s waybill with the "goods on order".
After processing the discrepancies (if any), the receiver makes necessary changes in the "goods on order" document according to the actual quantity, stock numbers, and prices of goods received and then he presses "accept" to convert "goods on order" document into "goods layout" forwarding it to the warehouse workers at the similarly named conveyor area.
Depending on how complicated the layout of a specific warehouse is more stages are possible but anyway the final one is "goods layout" turning into "goods received" when the replenishment cycle is over.
Please pay attention to the "conveyor" term.
Indeed, no information has popped out of nothing or disappeared to nowhere at any stage of the above chain.
After a sequence of transformations with the information (prices, suppliers, dates, etc.) being added to its specific content, the "procurement requirement" document created at the first stage finally turns into "goods received", representing the goods at the warehouse.
This is really similar to a motor works assembly line: first, we have a bare car body that passes through a strict succession of assembly sections, with new parts being added every step of the way.
This all results in a finished product the customer is willing to pay for.
The conveyor-like principle of the reflection of real-life business processes of the enterprise in the single system — as they exist in reality (rather than in the soaring fantasies of verbiage produced by all sorts of consultants) is already a giant leap forward as compared to the real practical impotence of modular ERP systems.
But this is just the beginning.
Let us return to our example of the "purchase" business process, and look at the people involved.
Starting from the category manager who creates the goods requirement list.
Why is a CM needed for it at all?
If we come to think of it, his work rules can be described by a (probably) long but generally straightforward algorithm: items with a sales history are purchased in accordance with their rate of sales. New goods are taken "on a trial basis", in varying quantities depending on the popularity of their category, price and consumer interest.
And that is all; we no longer need the costly (and likely corrupt) category manager: all his functionality is performed by a cold, error-free and tireless system far better than he can do it himself.
Then: the purchasing manager
Exactly the same, but the substituting algorithm is even simpler: buy where the price is lower (or the overall set of parameters is better: a decision to sign a contract of purchase takes into account the payment deferral time, rate of supply, warranty loyalty, etc.).
The purchasing manager is not needed, either: we set up a BTBX for suppliers and are sure to get better prices, now that the manifestations of human factors like theft and other faults are ruled out.
What about the prices? IEM System keeps the price points attractive automatically in the background mode.
If your business is not yet cool enough to keep suppliers frequenting your b2b floor, the managing system will form suppliers proposals on its own, by importing prices and inventories from Excel price lists (catalogues on websites, external online trading floors, etc.).
As a result, any job which can be described with a set of formal rules is fully automated by the reliable algorithms of IEM System.
If we return to the analogy with a motor works assembly line, the IEM System replaces the staff member attached to a conveyor section with a specialized robot whose managing software is based on a job description of its human predecessor.
A competent sceptic should ask: why not to program the same robots for conventional ERP systems and automate business processes in a way described above?
The fundamentals of ERP systems are the very obstacles.
Any conventional ERP system is a set of loosely coupled modules, in fact, being separate applications implementing their own functional fragments. A purchasing manager works within the SCM module, a warehouse — WMS, a call centre — CRM, manufacturers — MES, etc.
ERP systems break the procurement process being unified in real life (as any other) into several pieces according to the performing modules. These modules are interconnected with delayed data exchange procedures.
Each participant can only see the data of his module (usually not consistent with the data in the others), and real-time transactions are out of the question at all.
Operation-wise conventional ERP is a dumbbell and bottleneck: it is unable to provide participants in the value creation process with useful data in real time.
On the contrary, ERP is essentially an object for further manipulation that requires employees to type in data in addition to their primary duties.
On the contrary, the IEM System is an organizing core, a drum that sets a real-time pace that limits the level of freedom of the business processes being actually executed.
Apart from "robots" placed at different stages of the conveyor (i.e. automatically executing scripts for different stages of business processes), various scenario responses to arbitrary events can be formulated within IEM business logic inner space.
Each IEM robot, per se, is not especially smart.
But the combination of their high concentration in the operating system (thousands and tens of thousands of separate scenarios) and the oneness of the information field generate an effect that can be cybernetically close to the collective intelligence of social insects.
Same as ants: each ant, per se, is quite a primitive organism.
But as a group of many thousands controlled by a set of pheromones (way much simpler than the opportunity space of the IEM System managing actions and responses), they demonstrate quite a serious collective intelligence that can organize complicated societies and erect large-scale material objects.
Robots or automatic IEM scenarios can be compared with ganglia of individual insects.
Starting from a certain level of script complexity and coordination, the whole system acquires elements of intelligent behaviour in the context of response to changes in its external conditions (Swarm Intelligence).
Let us return to real business and to its main and rational IT task: to get rid of natural intellect bearers.
Seasoned by decades of marketers" monotonous chatter our reader will make a sceptical remark that not every position can be replaced with a set of automatically executing scenarios.
And not just many but most of them.
This is simply not true.
Let us recall a pre-requisite of the operational excellence of any business: the standardization of business processes.
At any enterprise, it ultimately results in dividing all the employees into three groups:
The ones whose job is described by a formal script
The ones not involved in value creation
While the third group mostly comprises useless personnel to be fired, the number of creative thinkers is usually much overstated due to insufficient depth of standardization.
Surely, poets and designers would remain creative thinkers.
But the others…
Being consistently used, a staff-replacing enterprise managing system reduces the administrative staff of a big enterprise (regardless of the area of its activity or business) from thousands to several dozens — or even just a few people.
But we haven’t discussed those very live people staffing the assembly lines, farm machine operators, warehouse labourers, and truck drivers yet. In other words — ordinary workers.
Technologies are advancing here, too; so factory workers will be replaced with cheaper robots, both tractor and long-haul truck drivers are already being replaced with autopilots, and some impressive gains have been achieved even in processing goods of different sizes at warehouses if we choose to believe Amazon demo videos.
As the respective technological solutions become widespread, they will be placed under the control of the unified IEM System to manage them the way it already manages warehouse WMS terminals or self-service checkouts.
The future is much nearer than we all think.