We have gone past the “Why do we need AI?” to “Why we should participate in building AI.” (As always "building" is both a noun and a verb.)
In my interview with Mehdi Nourbakhsh, What is AI?, we try to define "AI" for the purpose of this question and discussion session:
Mehdi Nourbakhsh - Many definitions of it exist, but I lean towards a scientific definition of AI. In 1981, scientists Avron Barr and Edward A. Feigenbaum defined it as follows:
“Artificial Intelligence (AI) is the part of computer science concerned with designing intelligent computer systems, that is, systems that exhibit characteristics we associate with intelligence in human behavior – understanding language, learning, reasoning, solving problems, and so on.”
I like this definition because it tells you that AI is a branch of science in which scientists design computer systems that mimic human intelligence.
To help you remember what AI is, I want you to think about it as a tree. This tree has several roots: philosophy, logic and mathematics, computation, cognitive science, biology, neuroscience, and evolution. The branches of this tree are sub-fields of AI, such as computer vision, machine learning, speech recognition, and synthesis, search and optimization, knowledge representation and reasoning, and robotics. Each of these branches has its own branch and sub-domain, or fields. For instance, the machine learning branch has reinforcement learning, supervised learning, and unsupervised learning as sub-branches.
Ken Sinclair - How can AI help the architecture, engineering, and construction—AEC—industry?
Mehdi Nourbakhsh - AEC projects are becoming more complex, and clients are demanding more. Our productivity across the AEC industry is low because of the lack of skilled labor and the bespoke nature of complex projects. Compared to the risks taken, companies have a low-profit margin because of bidding wars and the high cost of design and construction labor and material.
With the growth in population and changes in climate, we need to do more – a lot more. And we need to do it with very limited resources and in a very short period of time.
AI is not a silver bullet to solve all the problems ahead of us. But it is an underutilized technology that can help you to reduce the complexity of projects by running various scenarios and helping you make better decisions. It can help you improve the performance of a project either by giving you quick feedback about your design (e.g., the amount of embodied carbon) or helping you run your projects based on information. And, finally, it can help you to reduce costs. All of these things can happen while still fulfilling your obligations to your client and your company, and also have a positive environmental impact.
Ken Sinclair - How should AEC companies invest in AI?
Mehdi Nourbakhsh - You as the leaders of an AEC company might be asking yourself, “I’m working in the AEC industry, not in the tech industry; why should I have a data and AI investment strategy?” Before answering this question, I want to clarify that when I say you should invest in your data and AI, I don’t mean you should plan and build the next flying robot or autonomous bulldozer. Let the tech companies do that for you. What I mean is that you should have a strategy to find opportunities to use AI in areas where you want to gain or hold your competitive advantage in the market – for instance, if you have a secret sauce in estimating projects, designing buildings, or managing projects, and want to scale it across your company at your portfolio level.
On reading this interview, Elwin McKay-Smith, got inspired by the "Tree of AI" analogy in the AI interview and made this graphic of it.
Our June Issue of AutomatedBuildings.com, "Why AI?" was inspired by our Contributing Editor Sudha Jamthe, a Futurist and CEO of IoTDisruptions who mentors business leaders in AI and DataScience using No-Code AI, AI Ethics, and Capstone AI labs
She starts the answer to the question in her article Why AI?:
AI is everywhere and there is a growing rift between AI as a technology market and AI as a business or application market. AI is pervasive, everywhere inside our bodies, in our talking to us in our homes, bringing autonomy to our cars, watching us in city traffic lights, running factors and airports and monitoring our health, moods and predicting our behaviours and pushing us to shop, listen to music, exercise and live our lives.
1. AI is built by technologists and needs you
Technologies of the past were built and launched by technologists and worked fine. With AI they are built by technologists but are not scaling successfully because they need data, and in it is the call for your business acumen of what works for your business.
2. AI propagates biases from our imperfect world to our future
AI is biased and creates a feedback loop of leaving behind minorities who are not present in the AI building teams. Did you know that some top AI Technology models cannot recognize men with beards correctly? And of course, it feeds the bias of a few to target black people with recidivism and women from technology company jobs. So you need to be part of the AI building process to bring your lived truth to teach AI the world you want to build.
3. AI is creating our future and needs you to shape it
AI is trained by historic data. And in our history is good and bad what we have learned as humans. We need you to get involved in figuring out what from our collective pasts do we teach AI and what do we stop and change to make our future better for our children? It is all in the data that is used to train the AI. Also, AI is accelerating its learning to create original content -writing with new GPT-3, images with GAN (fake images), and non-existent synthetic voices. We all need to participate to shape this world to work for all of us to create the future we desire to create.
So when we think of “Why AI” let us think about our role in shaping AI at our work, in our lives, and into our future.
Explainable AI is the solution to make AI transparent and get all stakeholders involved to make AI transparent for all.
For a deeper dive into Explainable AI, check out this article from IBM Watson, What is explainable AI? Key passage:
Explainable artificial intelligence (XAI) is a set of processes and methods that allows human users to comprehend and trust the results and output created by machine learning algorithms. Explainable AI is used to describe an AI model, its expected impact, and potential biases. It helps characterize model accuracy, fairness, transparency, and outcomes in AI-powered decision-making. Explainable AI is crucial for an organization in building trust and confidence when putting AI models into production. AI explainability also helps an organization adopt a responsible approach to AI development.
As AI becomes more advanced, humans are challenged to comprehend and retrace how the algorithm came to a result. The whole calculation process is turned into what is commonly referred to as a “black box" that is impossible to interpret. These black-box models are created directly from the data. And, not even the engineers or data scientists who create the algorithm can understand or explain what exactly is happening inside them or how the AI algorithm arrived at a specific result.
In the article When AI and IoT Meet, Jason Pohl, Customer Success Manager for Buildings IoT adds:
Heightened concerns about climate change and rising energy costs have increased the need for building systems to take full advantage of artificial intelligence (AI) and Internet of Things (IoT) devices. While this process involves multiple layers of planning, system integration, and implementation of automation software, the result is a building (or buildings) that has clean insights into data patterns, increased analytics, lower energy costs, and increased sustainability. This also allows building operators to focus their time on more valuable work, instead of troubleshooting issues related to comfort
What is AI and how does it work with IoT devices?
The combination of IoT and AI better equips building operators to analyze and optimize their buildings’ energy efficiency data. Because of its incredible ability to uncover patterns in data, artificial intelligence encourages IoT devices to act intelligently with minimal human intervention.
When a connected device detects faults in sensors, equipment, or otherwise, AI learns patterns from that data. This learning capability is what makes machine learning (ML) a subset of AI.
However, it takes the right IoT equipment to build an integrated system that leverages the energy efficiency data patterns that artificial intelligence provides.
Keith E. Gipson, the CEO/CTO for facil.ai—not to mention longtime friend and industry pioneer—adds his thoughts in this interview, Why AI in buildings?
I grew up in this industry. It’s all I’ve ever done. I’ve used the tagline #automatedautomation for a couple of years now.
facil.ai Corp consistently saves 10-20% on the HVAC equipment operation, through reduced runtime with greater efficiency. Comfort is enhanced. Maintenance and truck rolls are reduced by up to 40%, in our experience. Our solution accomplishes this through the substitution of machine (Artificial) Intelligence instead of human intelligence.
It adds a level of both precision and accuracy to the underlying BAS. Goals are realigned. For example, there is no longer just a rudimentary PID loop trying at all costs to make a (sometimes hopeless) temperature set-point, wasting the maximum amount of energy possible by the way. But rather weather, occupancy, cost, and kW demand, among other things, are all taken into account.
I think Troy Harvey CEO of PassiveLogic said it best:
“Most Building Management Systems are packaged to look modern, but under the hood, they are the same old 1880s thermostatics, 1930s style proportional control systems, and 1970s procedural programming, Impossible to tune, and never optimized."
Human beings can’t scale and must be replaced by machines from the ongoing configuration and “tuning” process in building controls because we can’t help ourselves. We’re hopelessly biased. We also have at times, perverse incentives…
Eliminate the (human) bias and the perverse incentives. Put the “machines” back in charge. The automated driving car is an overused analogy, but I think in this case, it is very much appropriate. Imagine if a manufacturer came out with a “self-driving car” – yet the caveat was that the car can’t really “drive itself”; there must always be a human operator…just in case.
This is the current state of our, so-called, “automated” buildings. They’re not autonomous. They don’t function well without constant “tweaking” and human intervention.
My last question for Keith was, “There also seems to be some level of apprehension of putting AI completely in control of operating the building. Any final thoughts?” – and his answer gives a great summary for this month’s topic, "Why AI":
Some people fear AI. And perhaps they should fear it. Because if they’re part of the problem instead of the solution then they should be replaced. Not eliminated. Replaced.
Of course, there is a place for people to focus on higher-value work. Not “baby-sitting” the Pharmacy temperature or attempting to fine-tune (optimize) the Cooling Tower control strategy in a Central Plant.
Deborah Leff, Industry CTO of AI and Data Science at IBM said: “AI is not going to replace managers, but managers who use AI are going to replace those who don’t.”
This is upon us. AI is here in the Building Automation Controls industry, and it’s not going anywhere.