My first interaction with the term artificial intelligence, was when I saw the movie advertisements for Steven Spieldberg’s “A.I.”. In the movie, a robot named David, played by Haley Joel Osment becomes this near-future world’s first robot with human emotion and rationale. While androids are part of this world’s everyday life, they look exactly like humans but are programmed for specific functions and can’t go outside of their programming. Fast forward 22 years, and today I can’t have a conversation without AI coming up. But our AI is very different from the AI of early 2000s film.
Establishing some baseline definitions
I am starting with a disclaimer: I am no expert on artificial intelligence or related disciplines.
Some of my friends and colleagues work in this field, and there are many public experts who talk about AI ad nauseam. I will leave the technical opinions to them. This blog post will take you on my journey as I learn about AI. My goal is to build a basic foundation so that I can start formulating opinions about AI. Since my passion is combating climate change, and AI is fast becoming a part of everyday life, I will focus on how AI might affect the natural environment and humanity’s impact on it.
Artificial Intelligence and Machine Learning
Let’s start with some terminology.
Artificial Intelligence
“[Artificial intelligence (AI)] is the science and engineering of making intelligent machines, especially intelligent computer programs. It is related to the similar task of using computers to understand human intelligence, but AI does not have to confine itself to methods that are biologically observable.” – John McCarty (2007)
The field of artificial intelligence was founded as an academic discipline in 1956. The term has been around for longer, but the development of Dr. Alan Turing‘s Turing Test (1950) provided a basis for how we can define whether or not a machine is in fact “thinking”.
Machine Learning
You’ve probably heard of the term AI/ML. ML refers to machine learning, and is a sub-field of artificial intelligence and computer science. Machine learning uses data and algorithms to improve the accuracy of a solution. It is how computers can continue to learn and improve on their own and develop their own intelligence.
Some people use the terms AI and ML interchangeably, or together (AI/ML), but they are technically different. At least, this is my understanding from speaking with AI and ML experts.
Generative Artificial Intelligence
The last term I will define for you is generative AI (GenAI). This term was almost non-existent in the public realm one year ago. Now, generative AI is the type of AI most people probably think of when they hear or talk about artificial intelligence. Generative AI is a branch of AI which focuses on creating new content. Most recently, you might be familiar with ChatGPT or various forms of AI art. These are examples of generative AI.
If this is the first time you’re coming across some of these terms, that was a lot. To make things easier, I’ll abbreviate the definitions toward the end of the article below.
Artificial Intelligence in Sustainability
There are a lot of facets of sustainability in general which can and do incorporate AI. I’ll discuss a few examples at a high level, just touching on parts of the broader spectrum of applications.
Energy
Energy is a hot topic when it comes to climate change. In order to successfully combat climate change, we need to decarbonize now. But that is easier said than done, for multiple reasons. AI won’t solve all of our energy problems, but what can it potentially help with? I think a big area that AI is well primed for, is around preventative maintenance. A lot of time and expense is put into the maintenance of our energy systems. Newer technologies haven’t had decades to smooth out inefficiencies like our fossil fuel systems have. There are ways AI can help, including anticipating when components are reaching failure.
Renewable Energy
Another example of where AI is used is in renewable energy optimization. This is top of mind because it’s where my professional experience with AI started. AI and ML can be, and are, applied here for optimizing the performance of renewable energy assets, maintenance planning, and reporting. Some examples of companies that apply AI to wind energy production include: Onyx Insights; Clir Renewables; WindESCO; and Power Factors.
Grid Management
Another area within the energy sector that has a lot of potential synergy with AI, is grid management. Smart grids already merge multiple technologies for monitoring, control, and planning. They utilize electricity and data flow to improve energy generation, transmission, and distribution.
There are many solutions on the market today, from individual sensors to sophisticated management platforms. Solutions can help consumers understand where their energy is coming from. Or allow assets within the grid system to communicate to each other. Predictive control systems can help manage high-demand strains and distribute where energy draw comes from. AI can be applied to energy trading. Some examples of companies that apply AI to grid management include: Generac Grid Services; AutoGrid; and Origami Energy.
Food
As there are many facets of technology applications in energy, there are probably even more in food. The food and agriculture space is already being heavily impacted by climate change. Its continued innovation is necessary for the resiliency side of the equation of climate action.
AgTech
A popular area for research and development within agriculture, or AgTech, involves innovating plants and seeds. Another area is to look at the farming technologies and methodologies themselves.
Enko came onto my radar while reading the Gates Foundation 2022 Goalkeepers Report and listening to Bill Gates talk about ‘magic seeds’. I summarize the report more in a previous article. From what I can tell, in addition to seed innovation, Enko also has a technology platform. The platform focuses on protecting crops from pests and disease. They produce large discovery data sets, which are then used in their ML and AI models. Applications include site predictive analytics tools, decreasing design cycles, and finding safe, novel, and effective compounds quicker and cheaper.
Terramera is a local company focusing on the use of machine learning and artificial intelligence to support the growth of regenerative agriculture techniques, and turning agriculture into a carbon sink. Specifically, how does Terramera apply AI? They use it as a computational engine to power plant and soil health development. Terramera also has a technology platform, and work on reducing synthetic pesticides. Their two main products are Actigate and Miraterra.
Fishing
When we think of AI, we often think about super complex problems that humans haven’t solved yet. But in reality, AI has the potential to take very simple tasks, and simply do them a lot faster. OnDeck fisheries AI is a great example of this. The commercial fishing industry doesn’t have the best reputation when we talk about sustainability. But the UN estimates fish support the livelihoods of about 10-12% of people globally*, where seafood accounts for 17% of total human protein intake*.
Imagine you are on a commercial fishing vessel. You catch hundreds and thousands of fish a day. These fish are then counted and identified. Regulations (such as Canada’s At-Sea Observers Program) mandate commercial fishing operations have monitoring onboard their vessels. Such as mounted cameras recording their catches on video. The video is then sifted through back in the office. From my outsider perspective, this is an insane amount of manual labour. Increasing regulations and scrutinizing commercial fishing only makes this task more painful. Essentially, OnDeck’s product is a fish counter. It’s not hard for humans to do this work. But using AI they can help this industry become a lot more sustainable. If you’re interested, visit OnDeck’s website for a neat gif of their monitoring software in action.
Transportation
Transportation is another sector that contributes a lot to GHG emissions and climate change. In BC, our transportation sector accounts for about 37% of the province’s total greenhouse gas emissions*.
A few example applications of AI in the transportation sector include traffic management, self-driving cars, and fleet management.
For example, in Australia, they already have an automated speeding ticket system which prevents drivers from speeding. I recently read about a new adaptive traffic system being developed, called the Sydney Coordinated Adaptive Traffic System (SCATS).
Tesla first launched their Autopilot feature in 2015 and has since made autonomous vehicles a household concept. Like most of these topics, there is still a lot of work that needs to be done in this area. But I think it’s something I’ll see and experience in my lifetime.
Supply chain and shipping in general can, and in many ways already do, benefit from AI. As an example, I came across Samsara whose products focus on increasing safety, efficiency, and sustainability across a wide variety of industries. Including transportation and logistics.
Resiliency and Adaptation
Decarbonization, efficiency, innovating new ways to do old things. These only make up half of the equation when we talk about the climate crisis. There is no question the world has been permanently changed by human impact, and that we will keep seeing an increase of negative effects from climate change in the near future.
The term “Climate AI” has recently come to my attention, as a subset of Climate Tech. This area focuses on resiliency and adaptation around climate and weather. Where AI can be applied for predicting, mitigating risk and damage against, and increasing community preparedness against climate disasters.
“ML models are likely to be more accurate or less expensive than other models where: (1) there are plentiful data, but it is hard to model systems with traditional statistics, or (2) there are good models, but they are too computationally expensive to use in production.” – Tackling Climate Change with Machine Learning
Here’s an example application of AI for detection and community preparedness: Detecting Flooding in Fiji’s Croplands.
Open Ocean Robotics‘ autonomous boat is an example of recent innovation in climate data collection. Without sufficient data, Climate AI doesn’t work. In Open Ocean Robotics’ case, their utility is in oceanographic, meteoric, and current data collection, environmental monitoring, and real-time monitoring.
Sales Culture and AI
As someone who doesn’t work on AI directly, I often struggle to understand and then articulate AI applications. In one of my previous jobs, I worked for a company that incorporated AI and ML. Sales staff were taught to use canned one-liner responses. But some customers would want to get into the weeds. What part of the product is artificial intelligence? What about machine learning? Internally, there would be mixed responses from the data scientists and data engineers.
Some of the sales reps were hesitant to use the term AI or ML at all, because there wasn’t a good enough technical explanation for how AI was being applied. Other sales reps would throw buzzwords around whenever they thought it made sense. Then, when asked to elaborate, they’d talk about one of the more advanced capabilities in the product, regardless of its connection to AI.
I’ll always remember one of the conversations I had with a sales rep about AI. (this is a paraphrase) “We have to sell AI. It’s a requirement to even start a conversation with a company. All of our competitors do the same thing. That’s why you see AI and ML all over the place at conferences and industry exhibitions.” They continued, explaining that executives and directors who don’t understand what AI is, assume the product needs to incorporate AI. If it doesn’t, it’s not a very good product. I guess that’s why you often see AI and machine learning in marketing material for young technology companies. If they don’t have those words, investors won’t even begin to listen.
Positive outlooks
Of all the possibilities, where am I most interested in seeing AI progress?
If AI lets us quickly and cheaply model outcomes based on inputs, I see this as an important decision-making tool.
In terms of tackling climate change, I think about how climate models can help government develop policy. Taking environmental and economic factors in and making decisions with more clarity around the knock-on effects of one change (or lack of change).
I believe there are people working on this, though I don’t know much about what they look like or who is using their models yet.
In general terms, I am excited about AI automating the mundane. If 80% of our work can be given over to computers, and we are only required to do the last 20%, imagine the time we will have to spend on other stuff. Most of us are constantly consumed by our work, side projects, and gigs.
What if we didn’t have to work as hard to earn as much but the output was even better than it is now? If we could spend more time contributing to society, our families, friends, and our own health? How about those who now need multiple jobs to support a simple roof over their head and food on the table? What if they could get 80% of their time back? Possibly they would need to take on more work to live more comfortably, but they would still be working less than people who just have one full-time job now.
Cautions
Of course, the future of AI is not all positive. Only the naive would think that.
Humanity has shown time and time again, our big brains come up with a new technology and it soon gets out of our control and is integrated into ‘the system’ in which we live in. AI is and will be no different. Only AI has the potential to get out of our control much faster than any previously acquired technology. It could land us in a position of zero authority, for better, or for worse.
What do I worry about with AI?
I worry about exacerbating the already rampant imbalance in our labour market. I worry too many people will be left behind as technology skyrockets. What has stopped the most powerful people in our world from using their power irresponsibly in the past? Why would that change now with AI?
I’ve heard the analogy of comparing AI to the automobile. Our first cars didn’t have seatbelts. We had to build the car first, then develop safety systems after once we learned how the car would be adopted. I think that is an arrogant perspective that comes from people who have the privilege and power of being at the forefront of technology development. Where does morality come into the equation when we talk about AI? Hopefully, things like the EU AI Act will help buffer people from the worst potential outcomes.
Another area of concern that AI may help or hinder, is cybersecurity and digital identity. Fraud and identity theft, unfortunately, is not going away. And AI could advance protection against this, or help bad actors get even more sophisticated.
What can the everyday person do? You can sit back, wait and see what your future will bring. Or you can participate. Such as educating yourself, participating in forums, and using AI tools. Yes, you are giving your data away for free. But you are leveraging these tools to work for you. I can’t advise you one way or another. But I hope these notes from my knowledge journey will help you decide where to take yours.
Definitions
Artificial Intelligence (AI) – the science and engineering of making intelligent machines, especially intelligent computer programs.
Machine Learning (ML) – a sub-field of artificial intelligence and computer science that uses data and algorithms to improve the accuracy of a solution.
Neural network – a subset of machine learning machines can use to learn, based on passing data from one node to another connecting node. There are certain conditions that need to be met for data to be passed on, and different parts of the neural network can be weighted differently.
Unsupervised learning – uses data that is not labelled, and the models come to conclusions by looking for patterns in this data.
Supervised learning – needs the data to be labelled, or tagged. Each data input has an output label.
Predictive modelling – uses statistics to predict a future outcome or behaviour, based on historical and current data.
Adaptive control – is a control method that machines or programs can use when the system is dynamic. The controller can modify its behaviour based on the inputs and outputs of the whole system.
Generative AI – a branch of AI which focuses on creating new content.
More References
I’ve peppered specific references throughout this article, but here are some more general informative sources I’ve learned from recently.