Hack:
Let information itself be a "manager"
For corporate management roles, human resources are used to fill them. But what if the best ideas were allowed to manage also? Put an AI system in place that would test and prove ideas, selecting what would be the "best" and then letting that serve as a managerial guideline.
The focus on management and organization in a corporate setting tends to be based on using human resources. The problem is that a good idea may be buried in such an arrangement. And so it can happen that innovative breakthroughs may not see the light of day (e.g. commercialization) for literally years.
This problem is even more relevant now because Industrial Age management principles worked well in the past when managing matter (e.g. a widget factory) or energy (e.g. a steam engine) but do not work as well nowadays in the Information Age. That is because information is a totally different type of element (than matter or energy). It is much more volatile. It requires a different kind of management. Basically, you can manage information only by information. Thus, let information itself act as a "manager" of itself.
By using AI (artificial intelligence) to test and prove ideas (e.g. through thought experiments, logical analysis, prototype tracking, etc.), truly good ideas would have a better chance of survival. Thus, companies could better cash in on the creativity assets residing in their employees' brains.
So the above approach well allows for managing information in the only way it can be managed, namely, by information. So the organizational effort made here is really an organization of information itself rather than of human resources. Information itself serves the role of "manager."
This requires viewing the resources of a company in an entirely different light. It moves the organizational chart from a hierarchical construct (of people) to a networking construct (of data relationships). A modeling of such resources would be a foundational pillar to the AI system above-described. It could be domain-specific to the company (or division) in question. It could be started as a prototype at the scope of a small department and grow organically from there.
Good ideas would be allowed to surface much faster and have better traction as they are developed. Arguments for and against ideas could be thrashed out much more efficiently. For instance, if an additional point was suggested by someone to be added to what would be viewed (by the AI system) as a good idea, the system could process the potential almost instantly as based upon the context of the information model it is working with.
A greater chance of success and better quality of results could be had because the AI system had "proven' in advance that the idea "works." If the result is failure in real life, the system can be calibrated to learn from the experience and make sharper judgments. Of course, such systems are already used in various fields, but the uniqueness of this hack is in the application: The corporate setting of today that mimics the Industrial Age principles of the 19th century. It is for those entities that Gary Hamel describes in his book ("The Future of Management"), "Right now, your company has 21st-century Internet-enabled business processes and mid-20th-century management processes all built atop 19th-century management principles."
Adoption is probably going to be the biggest challenge because it is so radical in its approach. It completely disaggregates the organizational structure of Industrial Age management principles. One suggestion would be to slowly build human champions for it from within the organizational structure by having prototypes clearly provide strong benefits to the early adopters.
Another challenge will be in the degree of realism. An AI system is a poor substitute for the human brain which is one huge advantage to human resources (for organizational management). Also, the AI system would not be consciously aware of the decisions it is making. So the degree of "management" assigned the AI system will, by necessity, be limited in scope. One suggestion would be to limit the usage of the AI system to decision-making that is algorithm-based. In other words, if logic cannot be applied to a certain case area, have it be handled by human managers only. Also, a human override arrangement will ensure that the AI system is not the final authority.
Also, the AI system would, by necessity, have to be very sophisticated and thus very hard to produce. So the cost/benefit ratio would no doubt dictate the need for laser-like precision in determining the scope of the AI system's application. This is where prototyping is so important. "Easy" wins initially would indicate that the area covered allows for decision-making that is algorithm-based. Mountainous obstacles that appear early on can be a warning light for further endeavors (no progress at all may indicate decision-making cannot be algorithm-based).
Pick a department where the nature of its processes and responsibilities is very routine. Scope out a specific area of responsibility, then build an information model for that area, covering as much detail as practical. Then build a software prototype that can digest a specific decision-making task handled in that area, and then output the most plausible pathway to take. Test the results. Repeat as opportunity affords. Run an analysis on the results obtained, preferably against a baseline of historical human decision-making results for the same task, then compare effectiveness between the two (human and machine decision-making). If the AI system comes out better, then the best manager for that task is information itself.
Find out those areas in a company that have the greatest track record of repeated failure. Then filter out those situations that were clearly preventable (some issues resurface due to the "nature of the beast" and are unavoidable). Determine (if possible) which of those situations involve decision-making that can be algorithm-based. Of those situations filtered so far, pick which ones have the highest cost to the company (monetary or otherwise). Pick one of them to build an AI system prototype for, following the procedures given in the first paragraph above.
The timeless advantage of the web...I missed this one in 2013....thanks to your new effort in 2015 I had the chance to catch up.
I agree 100% about the necessity to look at company's resources in a new light as you propose.
By the way, 2015 is officially the year of the light by United Nations effort.
Thank you Georges for sharing thoughts in such a prospective hack.
- Log in to post comments
You need to register in order to submit a comment.