very interesting post you linked to. thanks for that.

I am somewhat with you on the idea of AI replacing humans, but I still would like to shoot for that as a goal. I have been reading and playing with AI for some 20+ years (remember The Fifth Generation?) and I have seen steady progress, but very, very slow progress.

I think we are going to end up inundated with data very soon (someone said in some article I read recently that Google and Yahoo process upwards of tens of petabytes daily — petabytes!) and we need to speed up the process of using the systems to help us manage the tsunami of data, information, and knowledge.

This is the goal of my writing, helping us see the problem of massive data inputs and no other way to process it than via massive computer power.

Else, we will not be able to make required decisions in real time as we want to do.

Thanks for the link and the read!

I really appreciate the way you frame the problem, and the solution, and I agree that storytellers are an interesting twist.

However, I cannot see replacing one with the other. I do agree that storytellers are a great way to provide a simple explanation to a complex topic, and I do favor storytelling over most anything else in this world when it comes to having those conversations — the power of parable cannot be replaced with most anything else. However, I don’t think that storytelling replaces decision making.

I think your question is key: as we talk about business, will CEOs trust systems to make the decisions? i think that the answer is it depends, they would if they can enter the logical constraints to bound the processing. That is my main argument here, and i still think that it is what scares us in this dialog.

Now, if we can figure out a way to take the stories told, unravel the logic boundaries, and use that as input — now we are talking! that would be an ideal world… humans can explain to the machines via a story what the decision making entails, and the machine can then make he right, rational, logical, decision based on those parameters. And humans can then use that decision to tell even better stories.

Maybe too far ahead…

Thanks for adding a great wrinkle to this conversation.

I am glad you see it my way

Seriously, I am glad you provide the link to the Lyzasoft WP, it actually is a step in the right direction in my opinion.

Maybe I am guilty in this debate of not providing a context to my comment (see, already failing at predating the logical processing by not providing sufficient parameters). Do I think this is a 2-3 year event? Absolutely not, we are talking at least 10-12 years, maybe even more than that.

So, why, as people would nicely ask, are we having this debate? Because, as you define in your WP and I have been saying, I am not about removing the human — I am about changing the order and having the human logic predate the processing and finding of the answer. It takes more than 12 months to build these systems, more like 12 years, so these are decisions we must make now: what is the role of the human in the hybrid model of the future?

You had some interesting ideas in your WP, and so have most of the other people who commented. Alas, why is it so hard to remove humans from decision making and simply let us tell the systems what the constraints are?

That is my advocacy point: let us become the logical constraints to the model, but not the model.

Thanks for a great comment!

I can sense that I am touching a nerve with complex transactions, but I don’t think it is because we don’t think that systems cannot create the right answer. I think it is because we are afraid that the complexity of the logical constraints we have to provide the systems to make the decisions are above what we can produce.

Are there decisions where human bias and sentiments should play a role? yes, definitely. But not in business. Biz is biz, not an emotional event. We need to be able to constrain complex problems with a specific set of logical parameters to allow the systems to make the decisions. We may not have the ability to do that, yet, but that does not mean the systems cannot do it.

I do believe, very strongly, that removing the human bias from business decisions will work better in the long run (few initial bumps at the beginning). And, no — it won’t turn into “Brazil” (the movie, not the country). The driving directives of the logical constraints shall (almost wrote will, bad choice of word) not allow for ‘evil’ to be the reason for the decision. If we can trust Google to not be evil, why not everyone else?

Thanks for the great conversation…

I agree with you on the power of analytics lying in helping us manage the reams of data and to spot patterns we would otherwise miss. But I am not entirely sold on the idea of sensemakers due to bias and lack of sufficient talent to implement the concept.

To me the power of automation lies in exactly what it does: give it marching orders (logical rules, bounds and parameters to constraint the processing) and let them go and do what they do best.

We are underestimating what we built, we can trust computers more than science-fiction writers tell us to. They can, indeed they do, provide the power to replace the most cumbersome tasks. I don’t think the humans are replaceable – the input i expressed above is always needed in analytics and processing – but I do think we are scared to test what we can do and what we can obtain from releasing raw processing power into our daily lives.

I am biased by a vision of automated processing and decision making where humans dictate the logic for the decision making. I think, as I said above, that removing the biases we inherently carry from the equation would make for better decisions. Sure, setting the constraints and parameters for operation may be hard and we may not be there yet. But instead of investing the time seeing how we can make sense of processed data after is is processed (and potentially requiring several passes to get a good result) — wouldn’t we be better off if we just let the systems do what they do better and we focus on what we should be doing better?

a perfect symbiotic relationship…

Thanks for the great conversation!

I am not sure i understand why certain topics would be proper and others may not. It sounds like we could not trust systems with certain complex matters? would their decisions be wrong? how does complexity affect sheer computing power? The systems provide the brute force, the human predates that force with the logic parameters. Complexity may men more logic parameters, but not necessarily more complexity for the system (probably more time).

One thing that has struck me through this (delayed due to my fault) conversation has been people’s reluctance to trust the system with complexity. I personally trusts us less with any decision where our biases enter the picture than I don’t trust “them”.

If you want to make the case that we may not know the entire logic parameters we need to use in complex systems, I can listen to that and somewhat agree. But it is out failing, and we need to work on that. Eventually we are going to be staring at both simple and complex decisions that will require extensive processing and will have to be handled by powerful cyber-entities.

We are falling behind if we cannot predate the power with logic. Systems can make sense if we provide the logical constraints, but they cannot operate without logical constraints.

Thanks for the read!

Thanks for the read, and glad to initiate the conversation at your site. I left you a comment, I think you are taking a very interesting position and something I have had experience with — where to place the human in the equation (similar to what I was telling Spiro).

To me the human predates the systems, making an symbiotic relationship — i left a couple of examples of models in your post.

Thanks
Esteban

It is a very interesting proposition, and I agree wholeheartedly that what we input into the systems is what will eventually produce the specific outputs we are seeking. Human filtering is done by the programming of the filters and ‘noise-reduction’ elements in the systems, there is no doubt. And it is not going away.

I don’t think that we can ever replace that layer, but I am very comfortable entering those filters before the processing of data and letting the system handle the rest on its own. I don’t need a sensemaker to interpret what the systems produced, I need someone to tell the systems how to interpret.

That is the single step reserved for the humans and one that – at least for now – cannot be replaced. It is obviously a matter of perspective, but I think that systems are ready to handle that load.

Thanks for the read!

Welcome to my humble abode — er blog. Thanks for the comment!

(yes, I am very late in replying — not intentionally just beyond busy writing more good stuff for you guys to tear apart — i mean comment on).

two parts: excellent question on what type of decisions. I am certain that we can find that just about any decision can eventually be made by a machine given sufficient parameters and data. The problem is not the type of decision to be made, but the availability of programmable rules and constraints to make that decision, or the existence of sufficient data to remove ambiguity (to a certain degree) from the decision. Similar as with human decision makers, at some point more data and rules won’t make a different and you have to pull the trigger and execute (I had this discussion with my wife maybe 50 times in the last two months alone!). I believe, and don’t have hard data to back it up yet, that these decisions are better made by machines without the human feelings of uncertainty, fear, and other emotions in the way. Not only that, but machines can execute without double-guessing. Would I trust anything to a machine today? no, not yet — but we are on our way to that. I am not going to say 200 years into the future, or even 100– my horizon is in the next 10-15 years. Before you refute it as impossible, keep in mind that 15 years ago the internet was only accessible via AOL and its friends, and only via dial-up (well, I had an ISDN line, but I was spoiled).

Second part, I like the concept of Sensemakers, don’t get me wrong, but I am very concerned with believing and relying on a human-machine interface to do the work. And, of course, here comes the trust/reputation portion again (as Mitch pointed out above). Yes, trust is a sine-qua-non requirement to rely on humans (machines see trust different, mostly as an outcome of good and / or bad programming, not as a matter of belief and reputation).

I find the idea of using sensemakers as intermediaries between machine output and as machine input very intriguing… crazy enough to work… it would also reduce the amount of work you have to do before the first analysis, and it fits in with years-old model of “secret customer service” where humans complement machines to provide better. faster, more accurate automated service (yes, automation is big with me).

thanks for an excellent comment, looking forward to more interactions.

trust and reputation are most certainly the achilles heel of the social world. the more social we become, the more that trust and reputation matters… and the more that computers can take over and analyze loads of information without bias. right?

at least that is the way i see it.