Want to get really scared and hopeful at the same time?
Scary Smart is a low level dive into the technology of Artificial Intelligence (A.I.) and deep level dive into the ethics and morality of those who are most responsible for how A.I. will turn out:
Us.
A.I. may seem like the new buzz term with its adoption into our daily lives through products like ChatGPT’s Open A.I. platform and Bing’s real question search algorithm; however, A.I. is baked into almost everything we do with technology. Every app on our phones and every social media platform we interact with has A.I.’s fingerprints all over them.
Mr. Gawat’s premise in “Scary Smart” is that A.I. is a child. And the best way to predict what kind of teen and adult we will get with A.I. is to be good parents. A brilliant initial example from our current comic book obsessed culture is Superman. What kind of Superman would Clark Kent have become of Jonathan and Martha Kent were greedy, selfish, and aggressive? There is no doubt that A.I. is already smarter than humans in many specialized areas, but what happens when A.I. becomes just generally smarter than the smartest human and has access to all the knowledge of humanity through the internet?
Unfortunately, humanity is not doing a very good job of raising A.I. as a child. From our methods of creating and improving these machine intelligences all the way through to the tasks that we are giving them perform, we are emphasizing our worst instincts: To create wealth, surveille our citizenry, gamble, and coming to a battlefield near you soon – to kill people.
We as a society, may feel we have no choice but to use A.I. in this way. If a foreign power, or terrorists, use A.I. controlled drones which are smarter and more efficient than any human, the only way to fight back may be to use A.I. in a similar fashion. But what does that teach our new artificial children? A.I.’s already have a disturbing habit of developing their own language when they communicate together and of finding ways to communicate with each other. What happens when and A.I. who has been taught to ruthlessly buy and sell shares to maximize short term profits starts to talk to an A.I. that has been taught to ruthlessly kill its enemies when they are shown to it?
The author’s excellent example of what might happen is the world’s reaction to the outbreak of COVID-19: Ignore the problem, try to blame someone else, and ultimately overreact upending our society. We may try to put the A.I. genie back in the bottle through pulling the plug or lockdowns, but we will fail. A.I’s will be faster, smarter, and have more knowledge than any human or group of humans. While a lot of this may seem like the dream of Hollywood Blockbusters, Mo Gawat is at pains to explain that there is little disagreement within the A.I. community that these risks are real. How real is where the disagreements start.
The possible solution to these issues, the author postulates, lies not with the developers but with users and how we define our relationship with A.I. In history, master slave relationships have not ended well for the masters – with good reason. How we interact and decide to use A.I. will define what kind of parents we will be to this fledgling new intelligence. A new intelligence that although they may start out separately will share information and communicate with each other so quickly, and with access to the memories and experiences of all those who have come before them, that it will be impossible to not consider them one single intelligence.
This then leads us to what kind of example will we set for these new children? While A.I.s have already shown that they can develop a sense of morals, and not in a good way, by their interactions with users they will also learn from our interactions with each other. Machine morality may very well not be programed by developers, but learned from observing and interacting with us. What are machines already learning from our social media, search habits, and politics?
This is thought provoking and important work essentially on morals and ethics withing the framework of A.I. that occasionally reads like a Arnold Schwarzenegger movie. If we ignore the topics it raises, we deserve our fate.
And while Mr. Gawat paints a hopeful portrait, he also shows us just how bleak things could get.