Chapter 17

IN THIS CHAPTER

Getting paid in space

Building cities in new locations

Enhancing human capabilities

Fixing our planet

When people view news about robots and other automation created by advances in technology, such as AI, they tend to see the negative more than the positive. For example, the article at https://www.theverge.com/2017/11/30/16719092/automation-robots-jobs-global-800-million-forecast states that using automation will cost between 400 million and 800 million jobs by 2030. It then goes on to tell how these jobs will disappear. A somewhat more measured article at https://mitsloan.mit.edu/ideas-made-to-matter/a-new-study-measures-actual-impact-robots-jobs-its-significant states that robots have cost us 400,000 so far, but it also states bluntly that robots are also lowering wages. The problem is that most of these articles are quite definite when it comes to job losses, but nebulous, at best, when speaking of job creation. The overall goal of this chapter is to clear away the hype, disinformation, and outright fear mongering with some better news.

This chapter looks at interesting new human occupations. But first, don’t assume that your job is on the line. (See Chapter 18 for just a few examples of AI-safe occupations.) Unless you’re involved in something mind-numbingly simple and extremely repetitive, an AI isn’t likely to replace you. Quite the contrary, you may find that an AI augments you, enabling you to derive more enjoyment from your occupation. Even so, after reading this chapter, you may just decide to get a little more education and some job training in some truly new and amazing occupation.

Some of the jobs noted in this chapter are a bit on the dangerous side, too. AI will also add a host of mundane applications to the list that you’ll perform in an office or perhaps even your home. These are the more interesting entries on the list, and you shouldn’t stop looking for that new job if an AI does manage to grab yours. The point is that humans have been in this place multiple times in our history — the most disruptive of which was the industrial revolution — and we’ve managed to continue to find things to do. If you get nothing else from this chapter, be aware that all the fear mongering in the world is just that: someone trying to make you afraid so that you’ll believe something that isn’t true.

Keeping Human Beings Popular

The headline for an online advertisement in the future reads, “Get the New and Improved Human for Your Business!” It’s one of those advertising gimmicks that many people find annoying. For one thing, something is either new or it’s improved, but it isn’t both. For another, aren’t humans simply humans? However, the headline does have merit. Humans are constantly evolving, constantly adapting to change. We’re the most amazing of species because we’re always doing the unexpected in order to survive. Part of the reason for this chapter is to ensure that people think about the future — that is, where we’re headed as a species, because we’re certainly going to evolve as AI generally invades every aspect of our lives.

Children (and many adults) love video games! For many people, video games are only so much wasted time, yet they have a profound effect on children (or anyone else playing them), as described at https://www.raisesmartkid.com/3-to-6-years-old/4-articles/34-the-good-and-bad-effects-of-video-games. In fact, playing games permanently changes the brain, as described at https://interestingengineering.com/playing-video-games-can-actually-change-the-brain. Video games are just one of many aspects of life that AI changes, and these changes aren’t generally appearing in new software, so the human of tomorrow is very unlikely to be mentally the same as the human of today. This likelihood leads to the idea that humans will remain popular and that AI won’t take over the world.

When you extend the effects of brain changes due to game playing, it’s not too difficult to assume that brain changes also occur for other uses of technology, especially technology that creates a Brain-Computer Interface (BCI), as described at https://hbr.org/2020/09/are-you-ready-for-tech-that-connects-to-your-brain. Currently, BCI enables people to do things like move limbs or overcome spinal cord injuries, but there is nothing preventing a BCI from letting humans interact directly with an AI in ways that we can’t even imagine yet. Far from being replaced by AI, humans are evolving to work with AI to perform amazing things that were never possible in the past.

Living and Working in Space

The media has filled people’s heads with this idea that we’ll somehow do things like explore the universe or fight major battles in space with aliens who have come to take over the planet. The problem is that most people wouldn’t know how to do either of those things. Yet, you can get a job with SpaceX today that involves some sort of space-oriented task (see https://www.spacex.com/careers/index.html). The list of potential job opportunities is huge (https://www.spacex.com/careers/index.html?department=), and many of them are internships so that you can get your feet wet before diving deeply into a career. Of course, you might expect them to be quite technical, but look down the list and you see a bit of everything — including a barista, at the time of this writing. The fact is that space-based careers will include everything that other careers include; you just have the opportunity to eventually work your way up into something more interesting.

Companies like SpaceX are also involved in providing their own educational opportunities and interacting with universities on the outside (https://www.spacex.com/internships/). Space represents a relatively new venture for humans, so everyone is starting at about the same level, in that everyone is learning something new. One of the most thrilling parts of entering a new area of human endeavor is that we haven’t done the things that we’re doing now, so there is a learning curve. You could find yourself in a position to make a really big contribution to the human race, but only if you’re willing to take on the challenge of discovering and taking the risks associated with doing something different.

Today, the opportunities to actually live and work in space are limited, but the opportunities will improve over time. Chapter 16 discusses all sorts of things that humans will do in space eventually, such as mining or performing research. Yes, we’ll eventually found cities in space after visiting other planets. Mars could become the next Earth. Many people have described Mars as potentially habitable (see https://www.planetary.org/articles/can-we-make-mars-earth-like-through-terraforming as an example) with the caveat that we’ll have to re-create the Mars magnetosphere (https://phys.org/news/2017-03-nasa-magnetic-shield-mars-atmosphere.html).

Some of the ideas that people are discussing about life in space today don’t seem feasible, but they’re quite serious about those ideas and, theoretically, they’re possible. For example, after the Mars magnetosphere is restored, it should be possible to terraform the planet to make it quite habitable. (Many articles exist on this topic; the one at https://futurism.com/nasa-were-going-to-try-and-make-oxygen-from-the-atmosphere-on-mars/ discusses how we could possibly provide an oxygen environment.) Some of these changes would happen automatically; others would require intervention from us. Imagine what being part of a terraforming team might be like. To make endeavors like this work, though, humans will rely heavily on AIs, which can actually see things that humans can’t and react in ways that humans can’t even imagine today. Humans and AIs will work together to reshape places like Mars to meet human needs. More important, these efforts will require huge numbers of people here on Earth, on the moon, in space, and on Mars. Coordination will be essential.

Creating Cities in Hostile Environments

As of this writing, Earth is currently host to 7.8 billion people (https://www.worldometers.info/world-population/), and that number will increase. Today the Earth will add about 156,000 people. In 2030, when NASA plans to attempt the first trip to Mars, the Earth will have about 8.5 billion people. In short, a lot of people inhabit Earth today, and there will be more of us tomorrow. Eventually, we’ll need to find other places to live. If nothing else, we’ll need more places to grow food. However, people also want to maintain some of the world’s wild places and set aside land for other purposes, too. Fortunately, AI can help us locate suitable places to build, discover ways to make the building process work, and maintain a suitable environment after a new place is available for use.

As AI and humans become more capable, some of the more hostile places to build become more accessible. Theoretically, we might eventually build habitats in a volcano, but there are certainly a few locations more ideal than that to build before then. The following sections look at just a few of the more interesting places that humans might eventually use as locations for cities. These new locations all provide advantages that humans have never had before — opportunities for us to expand our knowledge and ability to live in even more hostile places in the future.

Building cities in the ocean

There are multiple ways to build cities in the ocean. However, the two most popular ideas are building floating cities and building cities that sit on the ocean floor. In fact, a floating city is in the planning stages right now off the coast of Tahiti (https://www.dailymail.co.uk/sciencetech/article-4127954/Plans-world-s-floating-city-unveiled.html and https://www.greenprophet.com/2020/12/seasteading-floating-cities/ for an update). The goals for floating cities are many, but here are the more attainable:

· Protection from rising sea levels

· Opportunities to try new agricultural methods

· Growth of new fish-management techniques

· Creation of new kinds of government

People who live on the oceans in floating cities are seasteading (sort of like homesteading, except on the ocean). The initial cities will exist in relatively protected areas. Building on the open ocean is definitely feasible (oil platforms already rely on various kinds of AI to keep them stable and perform other tasks; see https://emerj.com/ai-sector-overviews/artificial-intelligence-in-oil-and-gas/ for details) but expensive.

Underwater cities are also quite feasible, and a number of underwater research labs currently exist (https://interestingengineering.com/7-things-you-should-know-about-the-future-of-underwater-cities). None of these research labs is in truly deep water, but even at 60 feet deep, they’re pretty far down. According to a number of sources, the technology exists to build larger cities, further down, but they’d require better monitoring. That’s where AI will likely come into play. The AI could monitor the underwater city from the surface and provide the safety features that such a city would require.

It’s important to consider that cities in the ocean might not look anything like cities on land. For example, some architects want to build an underwater city near Tokyo that will look like a giant spiral (https://constructionglobal.com/construction-projects/underwater-construction-concept-could-harness-seabed-energy-resources). This spiral could house up to 5,000 people. This particular city would sit at 16,400 feet below the ocean and rely on advanced technologies to provide things like power. It would be a full-fledged city, with labs, restaurants, and schools, for example.

No matter how people eventually move to the ocean, the move will require extensive use of AI. Some of this AI is already in the development stage (https://www.5gtechnologyworld.com/unlocking-the-mysteries-of-the-deep-sea-with-ai-enhanced-underwater-vehicles/) as companies develop underwater autonomous vehicles. As you can imagine, robots like these will be part of any underwater city development because they will perform various kinds of maintenance that would be outright impossible for humans to perform.

Creating space-based habitats

A space habitat differs from other forms of space station in that a space habitat is a permanent settlement. The reason to build a space habitat is to provide long-term accommodations for humans. The assumption is that a space habitat will provide a closed-loop environment, one in which people can exist without resupply indefinitely (or nearly so). Consequently, a space habitat would need air and water recycling, a method of growing food, and the means to perform other tasks that short-term space stations don’t provide. Although all space stations require an AI to monitor and tune conditions, the AI for a space habitat would be an order of magnitude (or greater) more complex.

Chapter 16 offers some discussion of space-based habitats in the “Taking your first space vacation” section of the chapter. Of course, short visits will be the first way in which people interact with space. A space vacation would certainly be interesting! However, a near-Earth vacation is different from a long-term habitat in deep space, which NASA will need if it actually succeeds in making a trip to Mars a reality. NASA has already commissioned six companies to start looking into the requirements for creating habitats in deep space (https://www.nasa.gov/press-release/nasa-selects-six-companies-to-develop-prototypes-concepts-for-deep-space-habitats and https://www.nasa.gov/feature/nasa-begins-testing-habitation-prototypes).

For some organizations, space-based habitats aren’t so much a means for enhancing exploration but rather for protecting civilization. At this moment, if a giant asteroid impacts Earth, most of humanity will perish. People on the International Space Station (ISS) might survive, however — at least, if the asteroid didn’t hit it as well. However, the ISS isn’t a long-term survival strategy for humans, and the number of people on the ISS at any given time is limited. So, people like the Lifeboat Foundation (https://lifeboat.com/ex/spacehabitats) are looking into space habitats as a means for ensuring humanity’s survival. Their first attempt at a space habitat is Ark I (https://lifeboat.com/ex/arki), which is designed for 1,000 permanent residents and up to 500 guests. Theoretically, the technology can work, but it will require a great deal of planning.

Another use for space habitats is as a generational ship, a kind of vessel to explore interstellar space using technologies we have available today (https://scienceline.org/2021/02/novel-science-talkin-bout-my-generation-ship/). People would live on this ship as it traveled to the stars. They’d have children in space in order to make long voyages feasible. The idea of generational ships isn’t new. They have appeared in both movies and books for years. The problem with a generational ship is that the ship would require a consistent number of people who are willing to work in each of the various trades needed to keep the ship moving. Even so, growing up knowing that you have an essential job waiting for you would be an interesting change from what humans have to deal with today.

Rather than build space habitat components on Earth and then move them into space, the current strategy is to mine the materials needed from asteroids and use space factories to produce the space habitats. The solar system’s main asteroid belt is currently estimated to contain enough material to build habitats containing the same area as 3,000 Earths. That’s a lot of human beings in space.

HABITATS VERSUS TERRAFORMING

Significant use of AI will occur no matter how we decide to live and work in space. The way we create the AI will differ depending on where we go and when. People currently have the idea that we could be living on Mars in a relatively short period. However, when reviewing sites such as https://phys.org/news/2017-03-future-space-colonization-terraforming-habitats.html, it becomes obvious that terraforming Mars will take a very long time indeed. Just to warm the planet (after we build the technology required to re-create the Mars magnetosphere) will take about a hundred years. Consequently, we don’t really have a choice between habitats and terraforming; habitats will come first, and we’ll likely use them extensively to make any plans we have for Mars work. Even so, the AI for both projects will be different, and seeing the sorts of problems that the AI will help address should be interesting.

Constructing moon-based resources

It’s not a matter of if we go back to the moon and build bases there; it’s when. Many of the current strategies for colonizing space depend on moon-based resources of various sorts, including the NASA effort to eventually send people to Mars. We don’t suffer from any lack of moon base designs, either. You can see a few of these designs at https://interestingengineering.com/8-interesting-moon-base-proposals-every-space-enthusiast-should-see.

At times, people have talked of military bases on the moon (http://www.todayifoundout.com/index.php/2017/01/project-horizon/), but the Outer Space Treaty, signed by 60 nations as a way to keep politics out of space (https://www.cfr.org/report/outer-space-treaty), has largely put an end to that idea. Moon-based structures and the services they provide will more likely answer exploration, mining, and factory needs at first, followed by complete cities. Even though these projects will likely rely on robots, they will still require humans to perform a wide range of tasks, including robot repair and robot management. Building bases on the moon will also require a host of new occupations that you won’t likely see as part of habitats or in scenarios that deal exclusively with working in space. For example, someone will have to deal with the aftermath of moonquakes (see https://www.nasa.gov/press-release/goddard/2019/moonquakes for details).

Using existing moon features to build housing is also a possibility. The recent discovery of moon structures suitable to colonization uses would make building bases on the moon easier. For example, you can read about a huge cave that’s suitable for colonization at http://time.com/4990676/moon-cave-base-lunar-colony-exploration/. In this case, Japan discovered what appears to be a lava tube that would protect colonists from a variety of environmental threats.

Making Humans More Efficient

An AI can make a human more efficient in lots of different ways. Most of the chapters in this book have some sort of example of a human relying on an AI to do things more efficiently. One of the more interesting chapters, though, is Chapter 7, which points out how an AI will help with medical needs in various ways. All these uses of an AI assume that a human remains in charge but uses the AI to become better at performing a task. For example, the da Vinci Surgical System doesn’t replace the surgeon; it simply makes the surgeon able to perform the task with greater ease and less potential for errors. A new occupation that goes along with this effort is a trainer who shows professionals how to use new tools that include an AI.

In the future, you should plan to see consultants whose only job is to find new ways to incorporate AIs into business processes to help people become more efficient. To some extent, this profession already exists, but the need will increase at some point when generic, configurable AIs become common. For many businesses, the key to profitability will hinge on finding the right AI to augment human workers so that workers can complete tasks without error and as quickly as possible. Think about these people as part script programmer/application packager, part salesperson, and part trainer all wrapped into one. You can see an example of this kind of thinking in the article at http://www.information-age.com/harness-ai-improve-workplace-efficiency-123469118/.

When dealing with human efficiency, you should think about areas in which an AI can excel. For example, an AI wouldn’t work well in a creative task, so you leave the creativity to a human. However, an AI does perform searches exceptionally well, so you might train a human to rely on an AI to perform search-related tasks while the human does something creative. Here are some ways in which you may see humans using an AI to become more efficient in the future:

· Hiring: Currently, a person hiring people for an organization may not know all the candidate’s real credentials and history. An AI could research candidates before an interview so that the hiring person has more information to use during the interview. In addition, because the AI would use the same search methodology for every candidate, the organization can ensure that each candidate is treated both fairly and equally.

· Scheduling: Today, a business is constantly at risk because someone didn’t think about the need to schedule a task. In fact, people might not have had time to even think about the need for the task in the first place. Secretaries and assistants used to manage schedules, but in the new, flattened hierarchies, these assistants have all but disappeared, and individual employees perform their own scheduling tasks. Thus, overworked employees often miss opportunities to help a business excel because they’re too busy managing a schedule. Coupling an AI with a human frees the human from actually performing the scheduling. Instead, the human can look ahead and see what will need to be scheduled. It’s a matter of focus: By focusing the human where the human can excel, the business gets more out of the human. The AI makes this focus on human excellence possible.

· Locating hidden information: More than ever today, businesses get blindsided by the competition because of hidden information. Information overload and ever growing science, technology, business, and societal complexity are at the root of the problem. Perhaps a new way to package goods exists that reduces costs significantly, or the structure of a business changes as a result of internal politics. Knowing what is available and what’s going on at all times is the only way that businesses can truly succeed, but the job is simply not feasible. If a human were to take the time required to become all-knowing about everything that a particular job requires, no time would be left to actually do the job.

AIs, however, are exceptional at finding things. By incorporating machine learning into the mix, a human could train an AI to look for precisely the right issues and requirements to keep a business afloat without wasting quite so much time in manual searches.

· Adaptive help: Anyone using products today will have to admit that having to remember how to perform a certain task is incredibly frustrating at times, especially when rediscovering how to perform the task requires using application help. You can already see how an AI becomes an adaptive aid when it comes to typing certain kinds of information into forms. However, an AI could go much further. By using machine learning techniques to discover patterns of use, an AI could eventually provide adaptive help that would help users get past hard-to-remember parts of an application. Because every user is different, an application that is hardwired to provide adaptive help would never work. Using machine learning enables people to customize the help system to fit each individual user.

· Adaptive learning: Today you can take an adaptive exam that tailors itself to ask questions about perceived weak areas in your knowledge. The adaptive exam either discovers that you really do know enough or asks enough questions to verify that you need more training. Eventually, applications will be able to sense how you use them and then provide automated training to make you better. For example, the application may discover that you could perform a task using five fewer clicks, so it could show you how to perform the task using this approach. By constantly training people to use the most efficient approach when interacting with computers or performing other tasks, the person becomes more efficient but the need for the human in that particular role remains.

Fixing Problems on a Planetary Scale

Regardless of whether you believe in global warming, think that pollution is a problem, or are concerned about overpopulation, the fact is that we have only one planet Earth, and it has problems. The weather is most definitely getting stranger; large areas are no longer useful because of pollution; and some areas of the world have, frankly, too many people. An out-of-control storm or forest fire doesn’t care what you think; the result is always the same: destruction of areas where humans live. The act of trying to cram too many people into too little space usually results in disease, crime, and other problems. The issues aren’t political or defined by personal beliefs. The issues are real, and AI can help solve them by helping knowledgeable people look for the right patterns. The following sections discuss planetary problems from the perspective of using an AI to see, understand, and potentially fix them. We’re not stating or implying any political or other kind of message.

Contemplating how the world works

Sensors monitor every aspect of the planet today. In fact, so much information exists that it’s amazing that anyone can collect all of it in one place, much less do anything with it. In addition, because of the interactions among various Earth environments, you can’t really know which facts have a causal effect on some other part of the environment. For example, it’s hard to know precisely how much wind patterns affect sea warming, which in turn affects currents that potentially produce storms. If humans actually understood all these various interactions, the weather report would be more accurate. Unfortunately, the weather report is usually sort of right — if you squint just right and hold your mouth a certain way. The fact that we accept this level of performance from the people who predict the weather testifies to our awareness of the difficulty of the task.

Over the years, weather prediction has become a lot more reliable. Part of the reason for this increase in reliability is all those sensors out there. The weather service has also created better weather models and amassed a much larger store of data to use for predictions. However, the overriding reason that the weather report is more accurate is the use of AI to handle the number crunching and look for identifiable patterns in the resulting data (see https://emerj.com/ai-sector-overviews/ai-for-weather-forecasting/ for details).

The weather is actually one of the better understood Earth processes. Consider the difficulty in forecasting earthquakes. The use of machine learning has made it more likely that scientists will know when an earthquake will happen (https://www.wired.co.uk/article/ai-predicting-earthquakes), but only time will tell whether the new information is actually useful. At one time, people thought that the weather could affect earthquakes, but this isn’t the case. On the other hand, earthquakes can affect the weather by changing the environmental conditions. Also, earthquakes and weather can combine to make a situation even worse (https://www.usatoday.com/story/news/nation/2015/05/02/kostigen-earthquake-weather/26649071/).

Even more difficult to predict are volcanic eruptions. At least NASA can now detect and obtain images of volcanic eruptions with great accuracy (https://www.livescience.com/58423-nasa-artificial-intelligence-captures-volcano-eruption.html). Volcanic eruptions often cause earthquakes, so knowing about one helps to predict the other (https://www.dw.com/en/volcanoes-and-earthquakes-the-pacific-ring-of-fire/a-36676363). Of course, volcanoes also affect the weather (https://www.sciencedaily.com/releases/2020/09/200911110809.htm).

The natural events that this section has covered so far are just the tip of the iceberg. If you’re getting the idea that Earth is so complex that no one person could ever understand it, you’re right. That’s why we need to create and train AIs to help humans do a better job of understanding how the world works. By creating this sort of knowledge, avoiding catastrophic events in the future may be possible, along with reducing the effects of certain manmade ills.

No matter what you’ve read, no way currently exists to prevent bad weather, earthquakes, or volcanoes. The best that humans can hope to achieve today is to predict these events and then act to reduce their impact. However, even the ability to reduce the impact of natural events is a major step forward. Before AI, humans were at the mercy of whatever event occurred because prediction was impossible before it was too late to truly act in a proactive manner to reduce the effects of the natural disaster.

Likewise, even though preventing all manmade disasters might seem possible, it often isn’t. No amount of planning will keep accidents from happening. This said, most human-made events are controllable and potentially preventable with the correct insights, which can be provided through the pattern matching that an AI can provide.

Locating potential sources of problems

With all the eyes in the sky today, you’d think that satellite data could provide an absolute source of data for predicting problems on earth. However, this viewpoint has a number of problems:

· The Earth is huge, so detecting a particular event means scouring millions of pictures every second of every day.

· The pictures must appear at the correct resolution to actually find an event.

· Using the right light filter is essential because some events become visible only in the right light.

· Weather can prevent the acquisition of certain types of images.

Even with all these problems, scientists and others use AI to scan through the pictures taken each day, looking for potential problems (https://www.cnet.com/news/descartes-labs-satellite-imagery-artificial-intelligence-geovisual-search/). However, the AI can show possible problem areas and perform analysis only when the images appear in the correct form. A human still has to determine whether the problem is real and needs to be addressed. For example, a major storm in the middle of the Pacific Ocean away from the transportation routes or any landmass probably won’t be considered a high-priority problem. The same storm over the top of a landmass is a cause for concern. Of course, when it comes to storms, detecting the storm before it becomes an issue is always better than trying to do something about it later.

Besides scanning images for potential problems, AI can also enhance images. The article at https://www.jdsupra.com/legalnews/artificial-intelligence-and-satellite-72364/ talks about how AI can increase the resolution and usability of images taken from space. By enhancing the images, the AI can make better determinations of specific kinds of events based on the event pattern (such as carbon tracking). Of course, if the AI hasn’t seen a particular pattern before, it still can’t make any sort of prediction. Humans will always need to check the AI and ensure that an event really is what the AI purports it to be.

Defining potential solutions

The solution to planetary problems depends on the problem. For example, with a storm, earthquake, or volcanic eruption, preventing the event isn’t even a consideration. The best that humans can hope to achieve today is to get the area of the event evacuated and provide people with another place to go. However, by knowing as much about the event as possible as far in advance as possible, people can act proactively rather than react to the event after total chaos breaks out.

Other events don’t necessarily require an evacuation. For example, with current technology and a bit of luck, people can reduce the effects of something like a forest fire. In fact, some fire professionals are now using AI to actually predict forest fires before they occur (https://www.ctvnews.ca/sci-tech/artificial-intelligence-can-better-predict-forest-fires-says-alberta-researcher-1.3542249). Using AI to enable people to see the problem and then create a solution for it based on historical data is feasible because humans have recorded so much information about these events in the past.

Using historical data to work through planetary problems is essential. Having just one potential solution is usually a bad idea. The best plans for solving a problem include several solutions, and an AI can help rank the potential solutions based on historical results. Of course, here again, a human may see something in the solutions that makes one option preferable to another. For example, a particular solution may not work because the resources aren’t available or the people involved don’t have the right training.

Seeing the effects of the solutions

Tracking the results of a particular solution means recording data in real time, analyzing it as quickly as possible, and then displaying the effects in a way that humans understand. An AI can gather data, analyze it, and provide several presentations of that data far faster than any human can do it. Humans are still setting the criteria for performing all these tasks and making the final decisions; the AI simply acts as a tool to enable the human to act in a reasonable amount of time.

In the future, some people might specialize in interacting with AIs to make them work with data better. Getting the right results often means knowing what question to ask and how to ask it. People today often get poor results from an AI because they aren’t familiar enough with how the AI works to ask reasonable questions of it.

Humans who assume that AIs think in a human-like manner are doomed to fail at getting good results from the AI. Unfortunately, that’s what our society promotes today. The Siri and Alexa commercials make the AI appear to be human, but it isn’t, of course. In an emergency, even with an AI accessible to the humans who are dealing with the event, the humans must know how to ask appropriate questions and in what way to ask them to get the required results. You can’t see the effect of a solution if you don’t know what to expect from the AI.

Trying again

The Earth is a complicated place. Various factors interact with other factors in ways that no one can anticipate. Consequently, the solution you created may not actually solve a problem. In fact, if you read the news very often, you find that many solutions don’t solve anything at all. Failure is the hallmark of many geniuses in the world, even technical writers, as described at http://blog.johnmuellerbooks.com/2013/04/26/defining-the-benefits-of-failure/. Trial and error help people understand what does and doesn’t work. However, by using an AI to recognize patterns of failure — those solutions that didn’t work, and why — you can reduce the number of solutions that you need to try to find one that works. In addition, an AI can look for similar scenarios for solutions that have worked in the past, sometimes saving time and effort in trying to find new solutions to try. AI isn’t a magic wand that you can wave to create a solution that works the first time you try it. The reason that humans will always remain in the picture is that only humans can see the results for what they are.

An AI is always programmed to win today. The “Understanding teaching orientation” sidebar in Chapter 13 discusses the potential for creating an AI that understands futility — that is, the no-win scenario. However, such an AI doesn’t currently and may never exist. Humans, however, do understand the no-win scenario and can therefore often create a less-than-optimal solution that works well enough. In assessing why a solution doesn’t work, considering the no-win scenario is essential because the AI will never present it to you.

The AIs you use in creating solutions will eventually run out of ideas, at which point the AI becomes basically useless. That’s because an AI isn’t creative. The patterns that an AI works with already exist. However, those patterns may not address a current need (one that you can see today, but haven’t creatively thought out), which means that you need new patterns. Humans are adept at creating new patterns to apply to problems. Consequently, trying again becomes essential as a means to create new patterns that an AI can then access and use to help a human remember something that worked in the past. In short, humans are an essential part of the problem-solving loop.