April 6, 2020
Cognitive Automation and AI in Business with Aera Technology and David Bray (CxOTalk)

Cognitive Automation and AI in Business with Aera Technology and David Bray (CxOTalk)

We’re exploring the concept of cognitive automation. I’m delighted to speak with two gentlemen
who will explain these concepts and what they mean for business. Fred, tell us briefly about Aera Technology. We build the technology that enables self-driving
enterprise. I’ll speak more about what it is, but it’s
fundamentally a cognitive operating system. I’m being a bit jargon-y here, but a system
that automates how decisions are made and executed primarily in large organizations. David, tell us about your background and what
you’re doing. My background is Executive Director for what’s
called the People-Centered Internet coalition. We strive to do projects that demonstrate
how the Internet can be used to make a measurable improvement in people’s lives. I’m also faculty at Singularity University
focusing on impact and disruption as well as Senior Fellow at what’s called the Institute
for Human Machine Cognition. David, automated cognition, what are we talking
about, human plus machine? Just set the landscape for us. The term “artificial intelligence,” which
right now has been used a lot in the last few years, we need to recognize the longer
history, that this is really the third wave of AI. The first wave of AI occurred about 40 to
50 years ago when Herb Simon was using it to demonstrate how machines could help solve
with games and they could actually solve those problems. Then about 20 to 30 years later, it was actually
then being used to actually solve what was called “expert systems” in 1970s and 1980s,
decision support systems. Then finally, now, in this third wave, we’re
looking at sort of the idea of neural networks, deep learning, the idea that what the machine
can really do is begin to sort of pattern match, what a human would typically do, at
a much higher volume and a much higher scale than would be possible if a human was to do
it by themselves. This then now gets to the idea of sort of
augmenting what a human does as a way of pairing the human with the machine so that the human
is learning from the machine and, at the same time, the machine is learning from the human
and, together, you’re getting better outcomes from them both. We’re moving from an era of people doing the
work supported by computer systems to an era where actually computers are doing a lot of
the work, the thinking that is required to make a decision and controlled by the humans. IT is becoming a reality, so people doing
the work supported by computers-to-computers doing the work controlled by people. That’s really the result of the acceleration
at which decisions have to be made, the increased level of accuracy, the increased level of
complexity that’s surrounding all the companies. They’re facing complex challenges and people
organized in a network are just not efficient enough to decide in real time with the right
level of accuracy. This is where cognitive augmentation comes
in again. Again, as David said, it’s a very interesting
process where the algorithms are getting enriched by the decisions that humans are making and
vice versa. We have a closed loop solution here. Fred, how is this different from computing
in the past, computing, doing work for people? Think about it as moving from an era where
users would log into a software and get the computers to actually process some computation–push
data up and down, forward, and so on and so forth–to having a system that’s completely
autonomous. We’re digitizing the process of driving. Well, here, that’s the same difference that
you see when you’re digitizing the decision-making process in an organization. You have a hands-free system that is actually
able to process the data, analyze it, come up with the recommendation, potentially go
to a user for control or, in some cases, automatically, in a touchless manner, execute that recommendation
or that decision. Moving from, as I said before, people getting
involved in every decision to actually a touchless planning, a touchless forecasting event or
re-optimization, all these kind of use cases that required people that now can be run completely
autonomously. David, this idea of autonomous, is that the
crucial dimension, the crucial element that makes these technologies and the impact different
from computing in the past? I think it’s a dimension. You can think about systems that are fully
autonomous. You can also think about some that are partially
autonomous. It’s also the dimension, though, of just the
sheer amount of data that you’re able to process now as a result of advances in terms of cloud
computing, edge computing that just were not possible in the past. The ability to process a massive amount of
data in real time at a reasonable cost is what really enabled the digitization of the
decision-making process. Computing infrastructure kind of set the stage
but it’s the data that brings it to life. Would that be an accurate way of describing
it? I’m going to make your analogy even a little
bit more complex. We’ve talked about two dimensions so far,
Michael. There is a third if you can imagine a 3D graph. The third dimension is the increasing, what
would be called, instrumentation of the planet with the Internet of Things, especially the
Industrial Internet of Things, although the commercial IoT is slowly picking up, especially
with voice-activated devices, with small satellites. We are now at a point in which the ability
to actually receive the data from the infrastructure is now conceivable where you can have this
augmented intelligence occur. It’s both the automation; it’s both the Internet-scale
computing power. Then it’s just simply the ability that we
are increasingly instrumenting the planet. Now, there are some cautions that come with
that. In some cases, there is a risk of becoming
either a surveillance state or surveillance capitalism, as some might say, if we’re not
careful. But this also allows organizations to begin
to actually be smarter about how they operate and have this augmented intelligence applied
to their processes. Fred, we’ve got this infrastructure and the
pieces with that; you’ve both just been describing. What then is the impact, the impact on people,
the impact on organizations, sustainability, even the environment and energy? Yeah, it’s a very good question. It’s very interesting. When we started that journey, we were really
exclusively looking at the impact on work. Oh, we’re going to be able to automate a lot
of decisions. We’re going to be able to augment the decision-making
capabilities on some very complex problems. You have to think about it in the context
of very large organizations that are being profoundly disrupted by the e-commerce giants,
as an example. We thought about the impact on the business,
on performance. We thought about the business on the work
and how people actually value their time at work. They sit inside this very large pyramid in
the case of a corporation, and part of their value is knowing how to work the system, how
to actually operationalize a decision and so on and so forth. All of that gets digitized. Where is the value left and how do you create
additional value for yourself? How do you monetize your subject matter expertise
and your time at work? That was the initial angle when we started
on that journey. But more and more, talking with our clients,
we realized that the impact is on the waste. We’re actually cutting waste in the entire
supply chain. We’re optimizing how decisions are made and
our trucks basically hit the road and, very pragmatically, how we’re consuming energy
and raw resources. The impact of that cognitive automation, enabling
companies to make better, faster decisions closer to the point of impact. To David’s point, there still need to be humans
in the mix. There are still some corner cases that are
not being properly addressed by data or by cognitive automation. But, in some cases, we can actually run better,
faster decisions with a massive impact on the environment, resource consumption and,
of course, on the way work is being done. The organization design, we can foresee in
the next few years that it’s going to fundamentally change from those very large organization
pyramidal structure to a network of smaller groups that will be tightly connected with
the ability to measure the impact of a decision on the different metrics in real time. I think you’ll see a deconstruction of the
large organizations, the way they’re actually structured. You’ll see an impact on the way people work
and you’ll get, of course, an impact on the environment in general. Fred, can you give us an example? You work with lots of customers. Can you give us an example of how this goes
beyond efficiency? Where does this human plus machine create
something that we could not do before? We have a platform on top of which we build
different skills. One of the skills we build is called a perfect
forecast. What it basically did is it’s now proven that
it’s delivering the right forecast for our client, which means what? If you know exactly what you need to manufacture
because that’s the right number that you’re going to sell, think about the savings across
the value chain from sourcing to manufacturing to the entire supply chain. Delivering the perfect forecast is actually
the key to a massive amount of savings. I’ll give you another example around promotion
planning. If you think about the way this process was
done in the past, people would build their promotion plan and roll them out once a year,
twice a year, and it would take a while for that to impact and hit the stores. But you have now consumers walking into stores
with their cell phone being able to check online coupons and their behavior, the way
they are actually consuming has completely changed. You have a complete disconnect between what
the organizations could do in terms of planning their promotions, which drives, in some cases,
50% of their sale, and the way the consumers are actually buying stuff today. Cognitive automation enabled these companies
to actually plan their promotions quasi-real time with end-to-end visibility in their supply
chain, understand the demand and matching the two. It’s their answer, so to speak, to the e-commerce
giants who have really built their success on incredibly sophisticated consumer analytics
and a very agile supply chain. David, it sounds like we’re not talking science
fiction here. There are actual use cases today of these
systems having a dramatic impact in many different areas. Yes. In fact, Fred’s example that he gave of the
shopping situation where maybe you’re either going to a grocery store or you’re going to
a clothing store, that’s something that’s only possible now and it’s happening now because
you do have sensors and devices in the store that are monitoring where the different customers
are going. Maybe you’re dwelling for an extra long time
at the vegetable aisle, and so then you could actually push to the customer specifically
and say, “Would you be interested if there is a special deal on broccoli or on lettuce?” whatever you’re actually looking at right
then and there. It’s targeted just to you and that’s only
possible because of the speed, that looks at your pattern of buying behaviors, looks
at what you’re interested in, and delivers it to you if you’ve given consent to receive
that targeted advertisement. The same thing for shopping. Another example that also Fred mentioned too
is supply chains. Up until now, supply chains were kind of something
that you had different sort of checkpoints along the way but you didn’t have real-time
visibility into the location and the timing of both things that you had and things that
you might need to have based on forecasts. You can actually begin to see how weather
might impact buying behavior, how weather might impact delivery behaviors. Again, this is the idea that what really is
happening is it’s augmenting the intelligence of the organization relative to how it engages
both its human assets as workers as well as how it interacts with humans as customers
such that it’s bringing together both Internet scale, assessments of data, it brings together
the sensors themselves that are bringing in this information, and then producing a result
that is not just about more efficiency but, also, about either more effectiveness or more
delivery of information or offering of services that are tailored to that individual. Again, this then raises questions in terms
of ethics, thinking about, “Well, when do we want our data used for this purpose? When do we want to actually have the sensors
being aware of what we’re doing and our buying behaviors?” These are huge questions to make sure that
we’re doing it with choice and consent, going forward, as opposed to people that may not
be aware of it and may not necessarily buy into having their data used for that purpose. Fred, as you talk with your customers, to
what extent do they appreciate or recognize the extent of the implications for the extent
of change that it may bring to their business as well as their industry and competitors? We talk to a lot of customers. The first thing I would say is that debate
that we’re having right now is a true C-level discussion. We’re engaging with the CIOs, the CEOs of
some of the largest companies in the world around that topic. I think they intuitively know that the way
they organize and the way decisions are being made in their organization is not sufficient
anymore. There is a drive for change. There is an impending event. When we launched Aera a couple of years ago,
it was like, “Are we out there?” The answer is no. The answer from these execs is more like,
“Where have you been? We’ve been waiting for a new set of tools.” The way decisions are made has not really
fundamentally evolved. We’ve got better collaboration tools. We’ve got better spreadsheets. We’ve got better planning tools that allow
us to compute faster, but the organization has not evolved. It’s the first time, with really the concept
of augmentation and automation, that we are seeing a leapfrog, a step change in the way
organizations are deciding on very simple and very pragmatic stuff on supply chain,
manufacturing, and the way they sell, as we discussed earlier. It’s going to change the business model and
the organizations very profoundly in the next few years. David, let’s shift gears a little here and
talk about the technology. We hear about machine learning. We hear about deep learning. We hear about neural nets, all kinds of different
terminology that I think most people don’t understand. Where does the technology fit into play into
this picture? How important is the technology relative to
all of these other pieces: the algorithms, the data, the computer hardware, and then,
eventually, the outcomes that a businessperson or a person in society experiences? Right. The reality is, the actual techniques that
are being used, whether it’s deep learning versus neural networks or something such as
that, that is actually less important than really three things. The first is the data and the data that’s
being used to sort of drive the automated decision-making. The old adage in computer science is, “Garbage
in, garbage out.” If the data doesn’t have the robustness and
diversity necessary to answer the questions as to whatever direction or process that you’re
trying to drive, then it’s not going to be sufficient. The second, though, is, of course, then the
sensors or how you’re acquiring the data. If the sensors either are missing something
or are not accurately pointing in the right direction or are not adequate enough to provide
the data that you need, then you will fall behind as well. Then, finally, it really is thinking about
how your organization changes how it operates. I think oftentimes we miss how old legacy
technologies can become a source of ossification for an organization not just because they’re
old and falling behind in terms of technology capability, but because organizations often
instantiate their processes in their legacy technologies. If that process itself needs to change, just
moving to a newer technology and not changing that process will pull the organization further
behind. What makes this really interesting, at the
end of the day, is the feedback loops that occur between the data, how you’re collecting
the data through your sensors, and how the organization itself responds as a result of
what the data is informing you to do next. If you have that cycle of feedback loops,
the actual implementations, the nice thing is you can rely on someone else to help make
that happen, but you’ve got to have those three things in key and then actually have
that quicker feedback loop so you can be responsive enough to adapt. Now, again, going back to the idea of this
exploration versus exploitation happening in organizations, I think what Fred said a
little bit earlier is absolutely right that the nature of how organizations themselves
are structured is fundamentally going to change; that we organized in the past with hierarchies. Hierarchies are absolutely the wrong thing
to have for this type of environment because they’re very good at efficiency and repetition
across the different organizational units, but that’s the last thing you want because
you actually want things to be fluid and adapt as necessary, which a hierarchy is not conducive
of. It’s really going to be interesting to see
where we go next with how organizations reshape themselves. The three pillars that he just described are
spot on. What I would say is, if you want to get to
a cognitive automation or augmentation, you need an end-to-end system. The algorithm that you just positioned in
your question are just a part of it. The closed loop that David described is exactly
right. Think about it as, again, you go back to my
self-driving car analogy. Having a sight of sensors and lighters and
GPS, all of that sitting there on the dashboard is not enough. You have to make it work together. You have to make it work in real time. The ability to actually process all this data
coming from inside the organization through the ERPs and the other transactional systems
and outside, the ability to process the data in real time so that you go to the users with
a proper recommendation to take an action, get the feedback from the users, and automate
the execution, that’s the problem that we’ve set ourselves to resolve, which is really
creating this end-to-end system. That’s why we call that a cognitive operating
system and not just a single piece of software. David, Fred was just talking about changes,
and you were as well, changes to organizations. This brings up this accountability and ethics
set of questions that have been kind of lurking in the background throughout this entire discussion. David, do you want to tee that up for us? Why accountability? Why ethics? Why is it so important today? Sure. I think, with the term “accountability,” that
can be a loaded term but it’s really saying, “If you’re going to start relying on this
interplay between data, sensors, and what the organization actually does and that feedback
loop, you have to think about it in terms of accountability, in terms of, one, who in
the organization is responsible for making sure what the machine is doing in an automated
fashion is appropriate and is right both for the company as well as for whatever customers
or members of the public that it’s interacting with.” There’s another pillar, though, too. The second pillar is thinking about if you’re
interacting with customers or members of the public, are they aware that they are either
giving their data to the sensors or are they aware that they are being serviced in this
fashion? Some may actually object to that or may have
qualms about that, and so it’s sort of the choice mechanisms that go with that. I think what you’re going to see, right now,
that the first forays with augmented intelligence are going to be in areas in which it’s a little
less controversial. It doesn’t actually begin to impede too much
into your lives, but it is things like supply chain behind the scenes to make sure what
gets delivered is what needs to get delivered. Finally, the third leg, as you look forward
to this, is really thinking about what is the future of human autonomy in the midst
of all this, whether you’re a human worker, whether you’re a human customer. Do you have any autonomy? Hopefully, you do. But what does that mean relative to all those
things that are happening in terms of automation and augmented intelligence? In terms of ethics, well, ethics is simply
the socially accepted, normative practices that we see that are appropriate here. We’ve seen changes before. The idea of privacy really is actually a 20th
Century idea that came about. It did not exist, say, in the 1600s or the
1700s. There may be other ethics that start to arise
that involve this and, ideally, thinking about what we can do to uplift as many people as
possible through what augmented intelligence provides. That, I think, is an obligation of CEOs, of
organizations, and of the public as a whole. What advice do you have for people in business,
for policymakers, for other constituency or stakeholder groups, which is basically all
of us? David, final thoughts on this? My answer is, they should have started paying
attention to these trends about a year or two ago because those companies that start
to invest right now in thinking about how they’re going to use data to drive decisions,
thinking about how they’re going to actually begin to instrument it, whether it’s their
supply chain, whether it’s what they’re doing in their stores. If you don’t put in place the investments
in the sensors, then you’re not going to be able to have that first mover advantage as
this rolls forward, and it is rolling forward now. The first idea is, start to be hungry, start
to explore this space, and see what’s out there. This is dramatic change in a very small period
of time. It’s not going to be like the 100 years that
happened with the Industrial Revolution. It’s going to be happening so fast that we’re
going to have to figure out ways to do both governance as well as organization in thinking
about what companies do and thinking about how communities do too. That’s going to have to happen in parallel
as we move forward together and it’s going to require looks across multiple sectors to
try and figure out the best way to move forward because not only will it be those companies
that do this first that will succeed; it will also be those countries that do this first
will have that first mover advantage as well. Finally, at the end of the day, it really
is about doing this to free the human to do more of the creative work, not the rote work,
not the repetitive work. We need to make sure, in the midst of this
whole conversation, we are talking about augmenting intelligence and making organizations operate
smartly. Think about what we can do to also embrace
that human spirit and uplift humans so they can either have more time to do more creative
work, more of the in-depth problem solving, more of the in-depth, “Why is this occurring?”
or also thinking about how they can give back to society in other ways as well, as we go
forward into this. It’s not going to be necessarily something
that means humans don’t still have value and purpose, but I think that’s going to be a
seismic shift because so much of our jobs right now will probably eventually be displaced
by this and we have to start having the conversation now about what does it mean to be human in
this era, have a sense of purpose, and actually be a member of society as well. Wow. That was a lot of good information, a lot
of insight. Unfortunately, we’re out of time. Fred Laluyaux, thank you so much, CEO of Aera
Technology. Thank you for taking time to be with us. Thank you very much, Michael. I really appreciate it. David Bray, Executive Director of People-Centered
Internet, thank you as well.

Leave a Reply

Your email address will not be published. Required fields are marked *