“The reality is, everyone of you today…you are a software company, you are a digital first company, you are building applications that are core…and at that core are digital products” — Satya Nadella, CEO of Microsoft — Ignite 2017
It doesn’t take more than a 90 second scan of your LinkedIn feed to notice that A.I. is all the talk on earnings calls on “the street”, a main topic of hope or despair in coffee chats within the halls of your IT or Marketing department, and has even proliferated from the dystopian block buster Matrix/Terminator schemas into our living rooms via complex sociocultural thrillers that challenge the very dynamics for which we “think” we are human (C’mon, please tell me you’ve watched “Westworld”?).
The reality is that the conversation about AI is increasingly important, so important, that Max Tegmark, a physicist at MIT and author of Life 3.0: Being Human in the Age of Artificial Intelligence, claims the conversation to be “the most important conversation of our time”. Max has a very interesting point, and for the leaders of organizations everywhere, you should be paying attention to what Max has to say if you care about how the success of your organization will be impacted by employees and technology interacting, because as you will see to Satya Nadella’s quote above, every business will become a software/data business (you can also reference my last article to see more about this topic). Currently Max’s thesis resides within the point of view that human beings have developed with a certain array of hardware (our bodies), for which we can “download” or update software to learn:
“We have enormous power to upload new “software” into our minds…if you decide you want to become a lawyer, you can go to law school, and law school involves uploading new algorithms into your brain so that now suddenly you can have the expertise of the lawyer. It’s this ability to design our own software rather than having to wait for evolution to give it to us that enable us to dominate this planet and create modern civilization and culture.”
Yes yes, I know where this conversation is going, how the heck does it have anything to do with running a P & L, tell me about how this relates to me!? According to Max’s work, he talks about the ability of software that is able to develop far smarter hardware or “brains” that can then remember and learn much faster than human beings ever could. “software that writes software”, is where these thought vignettes begin to escalate into how humans interact and utilize machine driven super intelligence to become ever more effective.
What Does “Super Intelligence” Have To Do With Business?
The Rise of Artificial Intelligence: Forbes
Everything. The success of any business relies on competition, and according to Michael Porter (think back to your Strategy classes and the 5 Forces), this doesn’t mean beating rivals, it means earning profits. The fastest way to drive profits is to increase productivity and performance within your current asset footprint, and cut costs and empower employees to get better insights faster (more about this below) which will also drive revenue and new incremental revenue generating streams. We are quickly approaching a period of time where our performance will be surpassed by computer performance, and businesses that embrace, build, educate, and deploy AI throughout their organizations will expedite their journey far faster than if they take the stance that “they aren’t ready for AI”. As an employee of Microsoft, I of course am biased on how magnificent our overall vision is in how we are partnering and stewarding with businesses to help them digitally transform — but our strategy is grounded within an elegant portal for businesses to progress and digitally transform.
An there it is again, “Digital Transformation”. We’re already far enough down the rabbit hole, so lets do this together: Digital Transformation is leveraging technology to create innovation and disruption that empowers employees and customers to achieve more. The four “pillars” in Microsoft’s “Digital Transfomation” value system of intelligence speaks to what this crazy AI conversation has to do with your business and may look something like:
- Truly engaging your customers by utilizing smart bots to gather real time intelligence on what your customers really value, and how you can better serve them with predictions on their potential to churn or opportunity to offer services/products that can better their lives.
- How you will empower your employees is not just a conversation about “how much money you spend on the Outlook’s or the Slack’s of the world”, but how much access you provision employees with to critical business indicators that can tell them how their P & L’s are performing in real time — dashboards can flash pretty pictures, but machine learning can tell you why charts are dipping or spiking, and can generate data visualizations for you to make better decisions NOW.
- Optimizing your operations is the only way for some businesses to stay afloat in the globalized market, no matter how good their products may be. Say you are a food manufacturing/distribution business, the three trends that are impacting your bottom line are 1) increased costs of commodities 2) increased regulations 3) larger competitor profiles and accelerating mergers amongst big businesses. Cutting costs by extrapolating data from the factory floor and having your data relate to each other, pulling out correlations between data streaming from critical applications, and understanding how each process is impacting each other will be absolutely necessary. Whether this is done through deep learning image recognition, intelligent edge devices that are connected to machines and are computing on the edge in real time, or by having machine learning models look at historical data and make recommendations- this is the NOW of competitive operations in a global market. If you don’t believe me, ever heard of the “Internet of OYSTERS”?
- Lastly, is the opportunity to transform products through AI First/mixed reality first applications: “In today’s world, my design team might make a mirror that we have to digitally render, send to the engineers so they can study it, and then make changes based on their feedback,” Wetzel says. “That takes time, and we find ourselves out of phase there a lot. But placing engineering and design in the same space, a process we call co-creation, streamlines that interaction.” — saysCraig Wetzel, Ford’s manager of design technical operations, as his team embraces Microsoft’s Halolens.
Unpredictable Artificial Intelligence: Easing Traffic Flow, Improving Corporate Strategy, and Possibly Tightening Marriages
Nicholas Christakis, a sociologist and physician at Yale University, ran a study within his campus sociology lab on the interactions between humans and AI. The study was composed of a game where humans and bots (simple AI programs) had to work together to make timed decisions. The net of the game revealed that humans tended to make patterned decisions, as did smart bots, which on average produced a similar decision scoring. Except for one group, which were bots that made randomized decisions 10% of the time — this group ended up creating more “noise” in the decision making process, which invoked different responsive thinking amongst humans, and led to faster decision making. This “dumb/guessing” AI allowed a randomized environment wherein humans flourished due to the “teaching” nature of bots as they performed a helper role in allowing “humans to help themselves”.
“Without the bot-added noise, people often got stuck in ruts”
So if computers that perform randomly can help humans make better decisions, and highly sophisticated AI intelligence can help humans get better at performing specific tasks (such as the AI entity “AlphaGo” helping human players play the game “Go” better), then why the big disparity between the desire for business leaders to deploy AI to human interactions within their teams, and the actual AI footprint that we are seeing in the worlds biggest businesses? Excellent question.
But utilizing dumb AI doesn’t mean just using smart machines to introduce random scenarios into a group setting to inspire thought creation, “One can imagine chat-bots mediating relationship therapy, by guiding couples toward compromise without succumbing to exasperation or boredom” says Dr. Cristakis. The article also references a recent study on how social media bots could “shame” derogatory tweets and lower the usage of racial slurs on twitter.
The “NOW WHAT?” of the AI Conversation
I thought a quick graphic here would help us before we go on any further. There’s been a lot of confusion as to how to tell the difference between deep learning, machine learning, and AI. This is an easy conversation to blurr, but understanding the differences is a key step in figuring out what is right for your business objectives.
Artificial Intelligence: Forbes
Typically, we see AI used as an umbrella term to cover the (3) circumstances outlined to the right, but lets get our hands dirty for a bit and get deeper. I think of “Project Malmo” when I think of Artificial Intelligence, which is part of Microsoft’s new approach to Research & Development. It involves an AI agent and a human player working together to build objects in a game called Minecraft. The human player teaches the AI agent how to use specific tools to reach the optimal level of the game, and then the AI agent teaches the human player lessons it has then learned on its own. Deep learning involves more of a multilayered neural system that has a refined ability to sequentially discern and discover as it becomes acquainted with new data — think of image recognition and all the things a machine needs to do to recognize the difference between a woman holding a surfboard and a red convertible.
Source: The Statistics Portal
The use cases I see being utilized the most in businesses today that are in the beginnings of digitizing, are deriving and automating from new insights using machine learning. They are using trained models to find patterns in uses case to detect things like why equipment is failing and how to predictively get ahead of future failures, and the actions necessary in real time to provision maintenance calls, intelligently procure parts, and equip technicians with the intelligence of the potential break down and which parts/tools will be needed — drastically lowering operating costs and decreasing down time. A good machine learning workflow usually looks something like this:
- Data Acquisition: finding relevant data to begin assessing variables you care about, and of course, making sure it is clean and ready for consumption
- Analysis: locating and understanding patterns manually, the foundation of data science
- Re-shaping: adding in functions and mathematical statistics to ensure correct categorization and to follow the flow of continuous iteration
- Modeling: create a model with the use of an engine, which is not a one time event, the process repeats over and over through a specific time box
- Refining: doing the above steps to increase the probability of accuracy, the more accurate the model, the better and more trustworthy the predictions
- Deployment: made available by applications, ready to be used by employees or customers
As we move from a mobile first cloud first technology ecosystem into an intelligent-cloud intelligent-edge ecosystem, businesses will now be able to discover new data environments, to discern patterns, and to compute and publish all at the very tip of user empowerment: through multi-sensors and multi-devices, at the “edge”.
As we can see from Gartner’s Hype Cycle above, we can try to get ahead of whether new technologies like AI are worthy investments now, or if we’re late to the game. Based on the chart, we’re right about at the time period where AI/ML/Cognition is seeing some big successes, but also a lot of failures. As we approach the next 5 years through the “Trough of Disillusionment”, it will be extremely important to make sure organizations align with AI providers that continue to evolve their mindsets, skillsets, and tools sets along with you, the business. It doesn’t just mean saying “we’ve done machine learning projects”, it is learning how many projects were successful, which projects failed, and what was learned from those failures.
So lets get going!…
While at Microsoft’s “Digital Transformation Academy”, a learning experience for Microsoft’s sales organization, I saw a compelling presentation by Capgemini (a partner specializing in helping businesses strategize around insights and data) on key components an A.I. project must have to be successful and what business leaders should be focusing on while they navigate the art of the possible:
…have a good understanding of the data that feeds these projects, Data & AI are just the means to an end — not the solution itself, and all AI projects require a culture that is willing to change and must be centered around trust, as this is the building block of interacting with Artificial Intelligences.
Originally posted at Towards Data Science