When you break it all down, without the ability to mass produce metal products (smelting); no skyscrapers, no airplanes, no automobiles, no, rockets, no computers, no Tesla….on and on. It’s the technology that’s fueled our modern world. So you ask, great, but what does a girder have to do with AI?
Everything. Let me explain.
Being a steelworker in the 70’s wasn’t just an occupation, it was a culture. Generations of steelworkers packed small towns all along the upper northeast, surrounding these steel smelts/factories. It was the lifeblood of every small town economy.
During the same time, mass globalization was also accelerating, forcing steel into a highly competitive and price sensitive commodity. Commoditization of steel put additional pressure on steel manufacturers that were constrained by powerful but antiquated unions with a nationalistic belief that the American Steelworker made better steel. This heavily restricted the ability of companies to reduce labor costs.
Knowing that something had to change, steel companies began introducing automation to the process. Automation was “packaged” as worker safety enhancements, while the real purpose was to cut labor/cost out of manufacturing. As the workforce was slowly cut, it became apparent to the unions what these corporations were doing, replacing workers with machines. This created hostility, and backlash that ultimately accelerated the demise of American steel. As it always does, inflexibility in self-protectionism will always lead to extinction.
Evolution is a process of adaptation driven by the pressures of the environment to not only stay alive but to thrive. When you look at the world of business, evolution has the same impact on corporations as it does on everything in our environment. With the introduction of automation, every decade has had to introduce automation to address market pressures.
As technology has become accessible to the masses, we have continued to see an accelerating pace of technological change where change itself, put’s pressures on markets, as well as the rapid change in customer expectations.
Let’s face it, we are in a continual state of exponential change. The fact technology evolves so quickly; it can frequently outpace for most, the ability to adapt to it. From services to micro-services, we are continually abstracting our code to try to adapt the exponential expansion in technological innovation. This is driving a new mindset – how you enable adaptation is more critical than the adaptation itself.
Within the enterprise software industry, the exponential shifts, are exceptionally problematic. Other than from acquisition, the established software architectures/stacks are challenging to evolve. This is due to both the development debt incurred to keep systems incrementally improving, and the cost of acquisition of newer companies/technologies. The desire for an enterprise to rethink its architecture only happens when an organization no longer can support its technology – both from infrastructure and customer expectation. Let’s be clear, gone are the days of “locked in” software license contracts or the acceptance of hard to use software. Startups continue to chip away at the enterprise because customers want more and want it easy to use.
In every company and software technology I have worked within; the focus has always been on making software easier to use for its users. We still approach everything with the human in mind. From role-based systems, to how we derive and display data, all constrained by permission-based or rule-based operations that are bound by human involvement and interpretation. Yes, we have some computational capabilities in our systems, but in essence, today’s software is just huge giant calculators.
With Artificial Intelligence/Machine Learning – the calculator model is entirely antiquated and wrong. The purpose of AI is not only to perform repetitive tasks quicker, more efficiently and more accurately, than a human, but to do so in a way where the AI, instead of humans, determines what it needs to do to attain the best outcomes. What makes this so compelling and unique, AI automation is only limited by the data in which it has access to. In sophisticated AI systems, AI intelligently sources new data to fill in gaps to not only improve its learning, but also to provide greater accuracy on its purpose.
Most notably, the autonomous vehicle is a direct result of a new AI-based architecture. Here we can directly see how automation completely removes the human from decisions (Also why Elon Musk is so concerned about the evolution of AI). Applying the same principle to enterprise software systems, you’ll soon see automation across data and business actions, automated sales teams, conversational marketing, all continually learning and improving through by using humans merely as data feeds. As these systems come online, you will start seeing areas of labor disruption inside of corporations, from customer service, accounting, HR, sales, etc.
With AI, it just needs to be “shown” its purpose, the purpose of what the algorithm is to achieve, a “Northstar.” From there the AI ultimately determines where it can and can’t find data to achieve the result, and when it comes to a specific area in which it needs clarity, this is where the human involvement takes place. (answering, suggesting, or entering.) In essence, moving from a human workflow/entry model to one of having simple prompt based conversations will be the norm. Just like human conversation, topics will change continually to better inform the AI, to provide more data to feed its purpose(s). What makes this so compelling, as the AI consumes new data, the ability to for it to ebb and flow happens real-time. Without this capability, things like autonomous vehicles could never happen.
The idea of creating a system based on conversation, itself requires us to blow up what we know of the software experience. While it seems we have very sophisticated systems today, from my perspective, we are at a point in the technological evolution timeline where we just moved from propeller to jet aircraft, and attaining a rocket won’t be in decades but in only in months.
These extensive AI systems require new architectures, not constrained by today’s human-focused approach. These new systems must enable dynamic and intelligent decisions that know “where” to get the information, from systems, data and humans.
Now, while this ultimately may be the future state of systems design, the reality is, in many cases, we cannot remove the human from this process as of yet. However, we should no longer design with a human focus. Instead, we need to consider the needs of the system and view human interactions as just another data feed.
From my perspective, the most important question to all of this is – what does this do to the UI? It changes everything. From task & workflows, prompts, requests, actions now all move and are simplified into a conversational UI. But this also includes looking at ways of integrating into voice systems, display systems, (surface computing) IOT, etc.
Amazon and Google and others are evolving their in-home devices (Alexa, Google Home). The integrated, conversational UI is still specifically only a “human initiated” system vs. a data initiated system. I view Alexa as a great demonstration of how my car voice command system should work. (I have never experienced a cars voice command system that was worth a damn).
Once we see Amazon/Google and others moving to a system initiated voice, vs. a voice-activated system, we will know that the shift to an AI-based architecture has occurred. It will open up the world of immersive experience. Again, it is not about the devices, but more around how the experience continues to be flexible to an ever evolving technological ecosystem.
We as technologists must rethink our approach to software, and how we approach it’s UI. Interlinking that conversational UI between data sources and human engagement is going to be key. It will no longer be about detailed workflows, but be based on questions and answers. The process of human/system conversation and full system automation is the fundamental shift and what will be the future of software. Overall, the blending of home/work device ecosystems and how quickly I can view/respond to relevant information (business, sales, relationship, etc.) where ever I am, will be its ultimate purpose.
I get that this is “big vision stuff” but this is an exponential shift in software architecture, its design, implementation and how it powers the operations of corporations. AI will streamline costs, automate processes and only use humans as a means to learn what it needs to make better decisions. In short, the business run by AI will shift from an operations focus to a data strategy focus, using human creativity to further the need to feed new data into the system continually.
Like the 70’s steel industry; automation will disrupt jobs. Industries will experience angry labor unions, and be self-protectionist. Just look at history. I counter though, instead of being inflexible, resistant or fearful, accepting that technology is changing everything we know, maybe its time for us to focus our efforts on what software can’t do; our human ingenuity and creativity.
Scary or not, it’s the future of business.