Team building insights from artificial intelligence
- 1. Team Building Insights from Artificial Intelligence Copyright © 2011 Robert Roan Robert.Roan@post.harvard.edu Because one of Artificial Intelligence’s main goals is to assemble results- oriented, autonomous thinking systems that are constantly getting better, it has a lot to tell us about the workings and needs of healthy, high-value teams. Where we may take certain things about intelligent behavior for granted, Artificial Intelligence (AI) researchers have needed to methodically deconstruct those assumptions in order to develop the thinking infrastructures for their systems. This article uses the Turing Test, which “is an essential concept in the philosophy of artificial intelligence 1” to develop a list of questions that will help you use tested scientific knowledge about intelligent behavior to construct your teams and their decision support systems. “In order to pass a well-designed Turing test, the machine must use natural language, reason, have knowledge and learn 2” Natural Language How will you develop and encourage active listening? “What is it you don’t understand about…” is not only a pop phrase but a testament to the challenge of understanding. How much time and energy have you seen wasted because of misunderstandings? AI has taught us that understanding language takes a lot of processing (brain) power so don’t expect that maintaining a shared and thorough understanding in a changing environment will be simple, easy, or one-time project. Knowledge Informed decision making should be as knowledge based as practically possible. Building your knowledge base ahead of time will make your decisions: 1 http://en.wikipedia.org/wiki/Turing_test 2 http://en.wikipedia.org/wiki/Turing_test
- 2. • Faster because you won’t have to spend time learning, and • Better informed because you won’t have to ignore knowledge that time pressures prevent you from getting. Some of the things to consider about the team’s knowledge base are: • The types of knowledge it needs, • How that knowledge will be acquired, • The form in which it should be stored, and • How the team will access it. The types of knowledge required for intelligent behavior include: How does the “world” work? This is a model of the domain in which the team will operate. Without an accurate model of cause and effect, it won’t know how to interact with that domain to achieve results. How do actions affect the world? You can’t reach a destination if you can’t put together a series of linked cause and effect steps. What actions are they capable of? It’s no use thinking about an advertising program that will cost ten times your budget. What’s the world’s current status? This includes objective information like market projections, price sheets, competitors, contact information, policies, procedures, etc. If you don’t know if you’re in New York or California, you don’t know what direction leads to Chicago. Knowledge acquisition involves both information gathering and transforming information into knowledge. From where will you get your initial information and knowledge? Most will probably come from your organization’s current knowledge and worldview plus research based on the same sense of opportunity that caused you to form the team. How will you augment that with additional information? How will you explore? Observe your best explorers and create a catalog of processes structured around knowledge needs or you may wind up with a lot of interesting, but irrelevant, knowledge.
- 3. How will you make sure that the information that presents itself during the team’s experience is not lost or forgotten in the journey toward results? Debrief and learn to mine indirection. Turning information into knowledge: How do you generalize from experience? How do you assess likelihoods? It’s extremely unlikely that there will be a guaranteed if-then result. Most things are probabilities. One way of thinking about the most appropriate container for knowledge is to consider where it fits in the explicit/tacit continuum. What knowledge belongs in databases, spreadsheets, word processing documents, presentations, etc? What knowledge will be on external web sites? What knowledge belongs in social media like wikis, shared bookmarks and blogs? What knowledge will be in stories? “…the most valuable type of knowledge is tacit knowledge, which cannot easily be codified or abstractly aggregated. Tacit knowledge, which often embodies subtle but critical insights about processes or nuances of relationships, is best communicated through stories and personal connections. 3” How will people find the information and knowledge they need when they need it? What will your librarian function look like? Reasoning Every decision will have ad hoc elements, but the reasoning process should be planned and practiced with as much support from systems and procedures as possible. This will make the process more • Efficient because you won’t go down (as many) dead ends, and 3 The 2009 Shift Index: Measuring the forces of long-term change. Deloitte 2009: https://www.deloitte.com/assets/Dcom- UnitedStates/Local%20Assets/Documents/us_tmt_ce_ShiftIndex_072109ecm.pdf
- 4. • Complete because you will have thought of what needs to be considered. What kinds of decisions will the team make? For each: What should trigger it? It could be the achievement of a milestone, an internal request from management or another group or an external event, such as the action of a competitor. What is the goal? If there are multiple goals, what are their relative priorities? What factors should be considered when making the decision? What knowledge and other resources will be required? How will the decision be implemented? Develop implementation competencies that can be easily invoked. For instance, how do you execute a price promotion? Identify the variables (product, price, duration. etc) and develop a system where you can feed in the variables and the promotion happens. After you understand the nature of the decisions: How will you train the team? They need to both feel comfortable with the process and know how to obtain the necessary resources when they need them. Learning “The most important thing to understand about the mind is that it's a learning device. 4” Every answer you give to the questions asked so far is, and will be, an assumption. How will you critique the validity of your understanding, knowledge and decision making? A coherent, structured process makes it easier to learn because you have a set of elements to analyze. Can you learn from failure? “What makes something worth knowing is organized around the concept of expectation failure. (Assumptions) are interesting not when they work but when they fail. 5” 4 http://edge.org/conversation/information-is-surprises
- 5. How will you make sure you learn the right lessons? The way you remember the world may change when the world didn’t. A hard time with a customer may harden someone when it was an exception and not a new rule. The world may have changed but your understanding didn’t. A feature set that used to be peripheral may become essential, e.g. a camera in a phone or free Wifi in coffee shops. How will you implement what you learn? Even if things meet your current performance standards, can they be done better? How will you intelligently explore the realm outside your comfort zone? Conclusion AI opens the door to do for cognitive processes what things like Lean Manufacturing and Six Sigma did for production processes. And in a world where thinking and imagining are critical core competencies for many companies, eliminating wasted and inefficient thinking may be the next competitive edge. Bibliography Russell, Stuart; Peter Norvig (2011-03-09). Artificial Intelligence: A Modern Approach Prentice Hall. Brockman, John. Third Culture: Beyond the Scientific Revolution 5 http://www.edge.org/documents/ThirdCulture/d-Contents.html