In part 1, we learned about recalcitrance and how it could prevent a fast takeoff of a superintelligence. We then drew a comparison to organizations investing effort into becoming smarter by learning and improving on a continual basis. How hard is it for an organization practicing continuous improvement to get exponentially better at continuous improvement? This depends on its recalcitrance, or tendencies to stubbornly resist improvement. An AI can be programmed to never give up when faced with adversity. People and organizations sometimes give up trying when their lean efforts face challenges, major distractions, or changes in leadership. Being aware in advance of the recalcitrance of our organization and its systems makes it less likely that we give up.
Organizations starting out on their lean journeys are nearly always biased to expect success. Lean transformation plans are optimistic. Its ambitions are high. One does not suggest a transformation unless the stakes are big. Successful lean efforts yield year-on-year double-digit improvements in safety, quality, speed and cost. In the first years when the pickings are easy, the improvement percentages in discrete projects can be embarrassingly high. So naturally the promoters and sponsors of lean put the benefits forward, often “locking them in” to annual business plans or budgets. Then they hold managers accountable for achieving these savings, using a methodology and philosophy that they barely understand. Some people may question the wisdom of the proposed approach, raise concerns about resistance to change, or point out how the lean model may not fit the business. They may try to quantify the recalcitrance of the organization. These are tough questions. Those who cask can be labeled “naysayers” and be seen as roadblocks in the road to easy double-digit gains.
How can we know what areas we need to work on if we want a fast takeoff, without a crash landing, for our lean transformation? Building on the ideas from Bostrom’s book, here are three ways to as whether the organization’s recalcitrance is greater than, roughly equal to, or less than the inputs of optimization power.
1. Studiousness. How studious are you? The smarts that we start out with is important. If we are you too ignorant to know that we are ignorant, as in the Dunning-Krueger effect, it can be hard to ever learn. We need to be self-aware enough to recognize blind spots. Studiousness comes from a combination of desire to learn, attention to areas needing improvement, and diligent effort and study.An organization that quickly tires from the effort it takes to learn has a higher recalcitrance. Like the tortoise and the hare, the organization willing to learn and change almost always outpaces the leaner-but-prideful organization over time.
2. Instrumentality. How is lean instrumental to the success of your organization? If the organization stopped practicing lean to focus on other projects or priorities, how would performance suffer?What specific capabilities does lean grant the organization? When intelligence is used to collect more resources and information, intelligence increases. What specific things does the organization aim its learning and practice towards? How do these things improve your intelligence and ability to solve problems for the organization and its customers? An organization that starts out with vague answers to these questions will fail to make a fast takeoff.
3. The box. What factors “box in” or prevent the organization from learning faster? Scholars propose that “boxing in” or cutting AI off from access to networks, processing power and data sources is the best way to prevent a runaway superintelligence. The boxing-in factors can be internal, such as physical limitations, or limitations in thinking and biases. We can learn our way out of these with will, effort and time. It is important to consider external factors. These include regulations that prevent changes or experimentation, market conditions that limit investment in learning, unreasonable customer requirements or unreliable suppliers that constrain efforts to improve. They can be funding sources tied to quarterly performance, or a lack of access to state-of-the art knowledge and practices. When boxed in by external factors, some organizations give up on lean, and some may even see it as a legitimate excuse to stop trying. Humans are resourceful enough to find ways out of the box if we ask the right questions.
How can our lean transformation achieve a fast takeoff? There are two ways. We can increase optimization power significantly or minimize recalcitrance. Investing more in the optimization power via training, coaching, workshops, and external consultants is expensive. The returns are limited not only by the quality of these inputs but also by recalcitrance. Directing efforts at the causes of recalcitrance is a cheaper way to get the maximum benefit from our efforts to learn.