Editors Note: This post is a part of a series on Ethics of Artificial Intelligence.
For the benefit of humanity, AI can prove to be the most important invention of our generation, if the strengths are appropriately governed. If you consider the innovation wave that AI is producing and the comparison between it and all previous generations, the opportunity for economic advancement, personal empowerment, invention, and maximization of humanity’s strengths has never seen a greater time of set before it. The benefits of applying AI to human challenges span from medical, to manufacturing, to safety, to personal well-being, to optimizing relationships. We’ll spend some time talking about the areas of life that will be positively impacted by AI, but these will only be a sample, because the most game-changing haven’t even been ideated yet.
- Low Value Work. The Digital Transformation era is often called the “fourth industrial revolution”, which is interesting because it stands to undo many of the negative results of the earlier industrial revolutions. In the first chapter we talked about the Human Difference, being the idea that humans have something which sets us apart from other animals or objects which infuses them with a greater degree of dignity. If this is true that humanity deserves dignity, then the best AI can produce is to enable that dignity in every person by optimizing the jobs that humans do vs. those by machines. A justifiable target of AI is to replace repetitive, low-value tasks with Weak AI algorithms, robotics, or interfaces which empower humanity to move into tasks which greater reflect the dignity of the person. In the book Les Miserables by Victor Hugo, the story of the galleys is told, where an entire industry was created to find young men, with very minor mis-deeds, and bring them into servitude in the galleys acting like machines, but as people being taken advantage of. Justice states that we should leverage all capabilities at our disposal to prevent people from being used as a means to an end, instead valuing them and helping them to be the best version of what they can be. AI has the opportunity to do just that if used to remove the need for humanity to perform tasks which could be automated by machines. The end of maximizing humanity will be only achieved however if the resulting world then assists the displaced individuals from the original jobs to do more dignified work, or to be paid a living wage from the work they could be doing. This we will discuss in a future chapter on humanity and re-framing the next steps.
- Medical Diagnosis. The opportunity to leverage AI for real-time medical diagnosis represents a stunning improvement in serving patient needs. A simple example is reading scans of the human body, where a Weak AI system has proven consistently to be more capable at diagnosing than a trained doctor. This is because the AI system gathers information from millions of previous examples so it can more precisely determine the actual problem than a human doctor with at most thousands. The machine simply can read and compare information faster, with more source material than a trained doctor can. This stands to optimize the ability for doctors to assist their patients by combining the power of AI with the additional control of a skilled human doctor. Think of how an engineer is more capable with a computer than a slide rule because he can do tension estimation for a large building. A doctor similarly would be more effective at serving a human patient with this sort of advanced asset.
- Medical Research. An area where AI is already paying dividends is in medical research, where studies which were once thought impossible or took many years can be completed much faster by leveraging the power of predictive analytics and AI. The more typical examples relate to simple statistics, where one set of scenario are compared with another in hopes of finding correlating learnings. As concepts like reinforcement learning enter the medical research market the opportunity to allow AI to self-research may provide additional advancements faster, since a researcher may be spending less time on the model. There is much confidence that in medical research we’ll continue to see an advancement in the leveraging of AI to bring about complex discoveries to help humanity thrive.
- Invention. The benefit area of invention has much in common with medical research, but I call it out because it stands for something a bit different. The capabilities of AI stand to help invention in a more simple way by executing experiments on ideas faster than ever before. This is where Weak AI will continue to be very effective and the inventions themselves may end up being based on AI itself. The creative ideation of inventions is the second area that represents the frontier of AI in this space. Think of an inventor trying to create a technology may in the future receive ideas from an AI platform, such as “did you think of?” or “what if we?” kinds of statements. We think of these as uniquely human, since computers don’t do them today, but what if they could look at permutations and make suggestions? That is where invention is not just statistics, but about looking at a picture and seeing a destination beyond what is easily seen.
- Fairness. The implementation of “fairness” in society is often debated, often complimented by the statement every parent has uttered, “life isn’t fair”. That said, what if a machine was able to help us to implement fairness in the way we operate as governments, businesses, and personally, if even to a degree? If when building a loan management platform the system ignores biases and ensures the right distribution is accomplished, or in scheduling workers, that the best timing is optimized for all employees? We know that people are often unfair and unable of making objective, unbiased decisions. A machine may be able to assist them in making decisions or implementing programs at large scale that have a fairness quotient and are cross-checked by a machine. In one sense fairness is an ethical area we need to watch out for with machines, but in another sense it represents a machine’s greatest strength.
- Environmental protection. It is clear by the many ways that AI is already being used to improve the environment that this opportunity has wide ranging impact. If you were to consider Microsoft’s AI-for-Earth program, which is positively impact everything from forest mapping, to species identification, to water conservation, you’ll see impressive benefits. The environmental protection research organizations were the first to leverage AI because much of their efforts relate to a statistical understanding of flora and fauna in certain locations. The expansion of AI-for-Earth, and those like it, has given the environmental community more real time capabilities to aid in improving the world in which we live.
- Safety. In perhaps an ironic twist for those concerned about AI’s negative impact, one of the most powerful areas AI can benefit humanity is in safety. The areas of safety that have a positive impact are wide reaching, including transportation, job site awareness, medical dosage dispersion, autonomous cars, and firearm identification. Think about an example where a manufacturing worker puts a dangerous piece of equipment on the ground and leaves it, instead of returning it to the staging area. An AI platform like Microsoft’s Cognitive Services could be used to identify the equipment on the ground, it’s location, and send someone (or a machine) to go pick it up and return it to its proper location. In another example, there is some concern about autonomous cars now, we’ll eventually see that they are in reality much more reliable and safer than a human driver. These, combined with thousands more, represent AI providing us a safer work and home environment.
- Personal Empowerment. The usage of AI for understanding our goals, unique talents, plans and how we’re engaging in those plans will prove to be invaluable. There isn’t a substitute for our human relationships in this area, but could serve as a very helpful complement. If you consider how we use tools already, such as Mint, to provide analysis of our personal financial position, you can understand how tools could provide prescriptive guidance in how we organize our money for future benefit. A similar capability could be provided on other personal goals, such as achieving a degree, tracking career progression, family time vs. work time, and personal scheduling. These are not limited to simple tracking (which we could do in a spreadsheet), but actual guidance or directional setting based on data and models aligned to your goals. This technology could even help individuals set goals, based on things that are important to them.
- Manufacturing optimization. The manufacturing area is driven by finding efficiencies for what it delivers, with AI being a chief area of finding improvements. If you consider an example of inventory optimization, which is the process of managing raw materials against built product and sales, achieving efficiencies here will net significant dividends. The combination of inventory optimization with location optimization, or manufacturing efficiency based on in-building layouts, the opportunity may be even higher. This type of enhancement drives real improvement that can immediate reduce waste and impact a company’s bottom line. The initial models in this space are driven by slowly testing historical models and arriving at ones which optimize the process based on known data. With the advent of techniques like reinforcement learning, AI may be able to impact these areas faster than ever before.
- Predictive maintenance. The often-cited use for AI in tool and machinery heavy environments, predictive maintenance has the goal of reducing downtime for the manufacturing or service process by mitigating risks of unplanned events. The capability of predictive maintenance is also helpful for companies selling products, as it allows them to serve their end-customers (such as a home consumer), but communicating a maintenance which would prevent an annoyance later, such as a home-well filling up with sludge, a furnace showing signs of failure right before winter, or a car that needs maintenance.
- Agricultural optimization. The agricultural industry has been surprisingly one of the first to see major benefit from AI based on targeted planting based on soil composition, weather study, and food storage. The farmer’s intuition may go a long way, but when complemented by data, especially when trying to increase yields without leveraging chemicals, data may be the greatest ally. In some cases entire planting operations are run by robotically engaged tractors which specifically target locations of seeds, when planting should happen, when watering is necessary, and when to harvest, even to micro-levels. This all enables farm-to-table by minimizing the time food is spent in transport and storage and maximizing the time spend growing and being available for a family to consume. It might strike in the face of tradition, but it may be the future of optimizing the family farm, at least in certain scenarios.
- Financial controls. The harms done to society by irresponsible and unethical management of financial markets and loan management cannot be understated. In fact, several of the world’s worst financial meltdowns have been caused by these very sources. The implementation of AI not just to optimize, but to protect the financial markets is often overlooked and could stand to help improve the predictability and control of the economic cycles. If we consider that how many act is motivated not by the common good, but by personal interest, we often do not have substantial controls to govern the financial environment for everyone’s benefit.
- Engineering modeling. The advent of “computer assisted design” was a big step forward in building and designing complex products, but not until recently has true AI started to be applied to this arena. The application of AI for stress modeling, material cost optimization, square footage design, customer experience, and financial zones changes this game completely. The ability for an engineer to leverage AI to bring an idea to life, then check it, vs. needing to do the detailed work could speed design from months to days, allowing for faster generation of what is needed, when it is needed.
- Education targeting. If any area will be necessary after the significant additional of AI, it is education, especially targeted education. We’ll speak about education and the AI market in detail later, but AI can itself help the education market by understand what people know, when they know it, and what they need to learn. This is where reinforcement learning can be very effective, by personalizing classes for a particular learner in an online learning environment and helping a learning to find the best path to their target knowledge. This does not replace in-person, classroom learning, but for many applications is it a very effective complement, especially in trades and technology education.
- Relationship optimization. Think about if you spend enough time with your kids? What about your spouse? What about your friends? What if you are looking for a spouse… where do you find them? This is what relationship optimization is about. If an AI system can help me to optimize my work and family schedule so I can spend more time with my family, or can simply remind me to focus on my family, it can provide a benefit. We already have millions of people who find their spouse online, but you can imagine those systems getting more and more sophisticated based on what it knows about us. That might sound creepy or strange, but it might be more attractive than hoping to meet that special person at a bar. We often think about technology as hurting our relationships, but in this case we may find that it actually could help us to focus on what is actually important.
- Home automation. There are strong opinions on home automation and the presence of AI, but with the arrival of technologies like Alexa, Google Home, and smart ‘things’, it’s rapidly approaching. If you ever left your lights on by accident, or the front door unlocked, or are concerned about safety, AI in the home can not only know about these things but potentially protect against them. As your pantry and kitchen are connected, your home may know you like peanut butter and toast with coffee in the morning and prepare it for you instead of you having to do it yourself. The robot vacuum may know it’s time to clean up based on sensors on the floor, or the garage door may open because you drive up the drive-way. These are all versions of applying AI or automation into the home sphere.
- Elderly protection. If you’ve ever worried about your loved ones living alone, with you, or in a group home, elderly protection, powered by AI may make sense to you. We desire the best for our parents and grandparents and sometimes taking care of them is more complicated than we might be able to do ourselves. The additional of elderly protective technologies, leveraging AI to enhance awareness of their situation, helps us be better caretakers. This should never be a replacement for the elderly being surrounded by caring, loving family members, friends, and helpers, but it might be a way to assist those individuals in being who they need to be for those we need to love.
- Public protection. In consideration of very visible situations like NSA-spying, governmental interference, and over-intrusion, this is a controversial area. We do know however that this type of protection has already been used to do very important things, such as the detection and analysis of child trafficking. It could also be used for detecting criminals with warrants in airports or public places, the detection of gunshots, or analysis of illegal drug distribution. The addition of AI to critical engagements like these has the capability to enhance the protection of the most vulnerable among us by reducing wasted time by law enforcement and directing them to where true problems exist.
- Power optimization. The industrial revolution was able to move beyond steam and other rudimentary capabilities because of the emergence of centralized power and its delivery into complex manufacturing and homes. Today we are more dependent on centralized power and distribution than ever, with it enabling much of what makes our home operate. The optimization of power production and distribution in its current form will serve to benefit substantially from AI in determining the quantity of production, where it is distributed, inventory management of fuel, and predicting build-outs based on growth patterns. The future of this technology is even more complicated as we see a macro-trend of distributed power, back to homes via smart roof systems and home batteries. This may not eliminate centralized power, but it will change the equation in a substantial way once the technology becomes approachable financially. The opportunity for AI to change this world and transform it is part of the way we will produce power cost effectively, efficiently, and environmentally consciously into the future.
- Precision exploration. The opportunities for precision exploration, especially in finding key resources, discoveries, or even emergency location of people is substantial. This type of technology is already being used today by the oil and gas industry, but its accuracy and the technology backing it will only improve. As AI becomes part of this, the predictions of individuals will be complemented by data from ML models that reduce environmental impact from drilling or help find discoveries faster. AI can enhance our ability to explore and see, such as finding a new species, the location of a sunken ship, or understanding tectonic plate movements. All these leverage AI to understand problems bigger than ourselves and aid us in making discoveries we could not make otherwise.
- Anomaly detection. We already have systems which perform anomaly detection today in some capacities based on simple tests of successful operation. The additional of AI to anomaly detection is to improve the detection of problems which don’t have a simple test, or to predict the coming of those that do. This type of technology can be very helpful in industries that produce composites of different materials, where the “cooking” and “mixing” of the materials needs to be managed and adjusted in near real-time. It is in these industries where intuition seems to play a large part due to those adjustments. The addition of AI to this picture may aid in the appropriate mix of components, the adjustments made real-time, and the outcome. The inputs based on data from the incoming components can aid a model in predicting a state that an individual would need to react to, though instead of reacting the “cooking” can be run properly from the start. This type of technology can also be useful in complex piping scenarios, where a leak needs to be detected before it causes catastrophic damage. The detection and location of the anomaly is important to the recovery process. In these and many more industries, anomaly detection will prove to be an important outcome of AI.
- Smart cities. The smart city is about bringing many of the separate AI technologies together, from protection, to optimization, to anomaly detection, to safety. A smart city is one that helps us maximize the best from every person by leveraging AI and connected technologies to enhance what people are capable of. A smart city helps us realize the skills and capabilities of every person through education. It helps us optimize efficiency through power optimization. It helps us build safely through engineering. This is where we bring technologies together to enable better lives for all people, not just a special few who can happen to afford a niche technology.
- Disabilities. In a very real and very beautiful outcome of AI we are helping those with disabilities, whether movement, sight, hearing, or understanding. Microsoft has created an example called Seeing AI, which enables individuals with visual impairment to hear the surrounding around them in audible description. For instance, an individual is staying at a hotel and needs to read their room number, this technology will identify the room # while they are walking and tell them. In cases of silence during a conversation it can tell the person of the faces of the people in the room. It can help read a menu or help them take a picture… as a blind person! These capabilities enable a level of independence for a person with a visual impairment and this is just the start for a whole industry. This is also a technology that stands to help not just those with physical impairments but also those with mental limitations as well. If there is a group that represents the Human Difference, that deserves our help and our love, this is it. This is where AI can help in transformative ways. A person who can hear for the first time, an individual with tremors who can write with a pen, identification of loved ones, helping people realize their worth and beauty. Let’s use AI to do that.
The areas we reviewed represent only a few of the areas where AI can positively impact the world around us, with each having many sub-categories of benefits to humanity. The addition of AI enables us to solve many of our greatest challenges faster and better than we ever could before. It is for the reasons we reviewed above and those we did not name, that AI is considered something we should pursue and will be pursued. For these areas to be successful, we must engage in the ethical problems with AI and mitigate them. The ethical problems must be addressed for us to receive the benefits without at the same time receiving even greater negatives that must be prevented. The opportunity is ours, and so is the responsibility.