Stamping out all chaos kills all the innovation too
Body
We need to tolerate the chaos innovation naturally brings, as long as it's controlled and managing. Stamping out all chaos robs us of our ability to meet customer needs in new ways.
Our competitors will meet customer needs if boxed-in thinking causes us not to
Body
Too much resistance to non-disruptive innovation causes an Aerospace company to give their competitor a leg up when they refuse to meet a customer need.
Focusing on consistent growth means allocating investment wisely using a solid business case process
Body
Pursuing consistent growth means investing in those opportunities most likely to produce it. Companies that don't collect or evaluate business cases can't do that. Those seeking to stand up a business case process encounter a small set of common avoidable pitfalls. Once they have a business case process that separates low risk from high risk efforts, they can evolve to a process that separates high risk innovative efforts from high risk bad ideas.
Companies with rigid or stagnant compliance policies won't tolerate or manage the disruption innovation naturally causes
Body
Companies aren't ready to seek nonlinear growth or respond to disruption in their markets if they have a rigid approach to their existing policies. Companies that use existing internal policies as an excuse not to change stifle innovation and ironically expose themselves to more (not less) compliance risk. Good internal policy requires constant review and tailoring and is a prerequisite for managing the disruption that innovation will require.
AI helps us address novel problems in new ways. But it tempts us to use it for everything--even things it's not good at. AI is nondeterministic--it gives different answers at different times even when we keep our inputs the same. If we want standardization, as we do in DevOps pipelines, there's a way AI can get that wrong.
The data is just the data; when we regard it as "good" or "bad", we miss our chance to innovate
Body
While we're right to create simple impressions of the data our organizations encounter, we miss the ability to innovate whenever we regard data or people that contradict our model as "bad".
Companies grow incrementally when they demand expenditure be justified through business cases, which predict near-term cash inflows resulting from the expenditure with high certainty. When they do, companies eliminate expenditures unsure of their future cash results, which typically include: 1) bad ideas and 2) critical new ways of doing things with hard-to-predict return timelines. To achieve breakout growth, companies need a business case process that allows innovative ideas while still deprioritizing the bad ideas. Most don't.
AI inference imposes substantial compute costs, which are not sustainable at certain stages of the AI development process. Several techniques exist to reduce compute requirements and associated costs. Here we detail decoupled pipeline stages.
Companies grow nonlinearly and respond to disruption primarily through innovative efforts. Innovative efforts require diverse resources--primarily talent and various types of funding. An innovative effort lacking any one required resource will fail. When companies divide management of those required resources among various managers, any one manager can stop any particular innovative effort. And if all innovative efforts receive all but one necessary resource, even if each effort fails to obtain something different, no innovation will take place.
