I love this format of an ongoing intellectual conversation. I don’t mind the wait between responses either because you are both real professionals in this area, not full-time social media influencers. That makes each response more of an investment with more value attached to it. Keep it up!
Well said Lena! Your discussion of how our cognitive autonomy is already constrained by myriad internal and external factors, not directly related to AI or autonomous systems, is excellent.
Along related lines, there is a belief that more AI and autonomy will remove crucial decision space from humans. It's well founded, and Brad laid out his case in ways that make a lot of sense. Yet I take the view that just as AI is considered magic until it's merely another app or embedded software, humans are capable of adapting very well to a world of increasing autonomy.
As you noted, this will require, of course, much more attention on human-machine integration. Delineating roles, responsibilities, dependencies, and interdependencies between humans and 'smart' machines. If we don't get that right, then Brad's scenarios become increasingly probable.
Francois Chollet's posting on the differences between cognitive automation, cognitive assistance, and cognitive autonomy is one of the best summaries I've read. I agree with his argument that what we need is more cognitive assistance, not cognitive autonomy (and it's also the more likely future).
I love this format of an ongoing intellectual conversation. I don’t mind the wait between responses either because you are both real professionals in this area, not full-time social media influencers. That makes each response more of an investment with more value attached to it. Keep it up!
Thanks for following along and engaging with it, Lane!
Well said Lena! Your discussion of how our cognitive autonomy is already constrained by myriad internal and external factors, not directly related to AI or autonomous systems, is excellent.
Along related lines, there is a belief that more AI and autonomy will remove crucial decision space from humans. It's well founded, and Brad laid out his case in ways that make a lot of sense. Yet I take the view that just as AI is considered magic until it's merely another app or embedded software, humans are capable of adapting very well to a world of increasing autonomy.
As you noted, this will require, of course, much more attention on human-machine integration. Delineating roles, responsibilities, dependencies, and interdependencies between humans and 'smart' machines. If we don't get that right, then Brad's scenarios become increasingly probable.
Francois Chollet's posting on the differences between cognitive automation, cognitive assistance, and cognitive autonomy is one of the best summaries I've read. I agree with his argument that what we need is more cognitive assistance, not cognitive autonomy (and it's also the more likely future).
https://fchollet.substack.com/p/ai-is-cognitive-automation-not-cognitive