Artificial Intelligence (AI): Benefits & Risks


What is Artificial Intelligence (AI)?

Artificial Intelligence (AI) is the field of computing that simulates the human learning process using hardware and software to do so. AI or AI (Artificial Intelligence) is currently considered a very promising field, as it allows human-like intelligence to be simulated.

Among the various information available for research on the internet, the Wikipedia page has a very interesting definition of AI, as shown in the link below.

How does AI work?

From SIRI to self-driving cars, Artificial Intelligence (AI) is making rapid progress. While science fiction often portrays AI as robots with human characteristics, AI can encompass anything from Google's search algorithms to IBM's Watson to autonomous weapons.

With the advent of Big Data happening, we see that the use of AI will increasingly help humanity to solve complex problems and activities that would be tedious or dangerous to human beings.

Artificial Intelligence today is known as AI (or weak AI) in that it is designed to perform a narrow task (e.g. just facial recognition or just internet searches or just driving a car). However, the long-term goal of many researchers is to create general AI (AGI or strong AI). While narrow AI can outperform humans on any specific task, like playing chess or solving equations, AGI would outperform humans on almost every cognitive task.


In the short term, the aim of maintaining the impact of AI on society is beneficial, motivating research in many areas, from economics and law to technical topics such as verification, validity, security and control. Considering that it can be little more than a minor annoyance if your laptop fails or gets hacked, it becomes even more important for an AI system to do what you want it to do if it controls your car, your plane, your trading system. automatic or your electrical network. Another short-term challenge is to stop a devastating arms race in lethal autonomous weapons.

In the long term, an important question is what will happen if the pursuit of strong AI is successful and an AI system becomes better than humans at all cognitive tasks. As pointed out by IJ Bom in 1965, designing intelligent AI systems is itself a cognitive task. Such a system could undergo recursive self-improvement, triggering an intelligence explosion leaving human intellect far behind. By inventing revolutionary new technologies, this super-intelligence can help us eradicate war, disease and poverty, and therefore the creation of strong AI could be the greatest event in human history. Some experts have expressed concern, however, that it could also be the latter unless we learn to align AI's goals with our own before it becomes super-intelligent.

There are some who question whether strong AI will ever be achieved, and others who insist that creating super-intelligent intelligence is sure to be beneficial. At FLI, we recognize these two possibilities, but we also recognize the potential for an artificial intelligence system to intentionally or unintentionally do great harm. We believe that research today will help us prepare for and prevent such potentially negative consequences in the future, thus harnessing the benefits of AI while avoiding pitfalls.

HOW DANGEROUS Artificial Intelligence CAN BE?

Most researchers agree that a super-intelligent AI is unlikely to display human emotions such as love or hate, and that there is no reason to expect the AI ​​to become intentionally benevolent or malevolent. Instead, when considering how AI could become a risk, experts think two scenarios are very likely:

1. AI is programmed to do something devastating: Autonomous weapons are artificial intelligence systems programmed to kill. In the hands of the wrong person, these weapons could easily cause mass deaths. Furthermore, an AI arms race could inadvertently lead to an AI war that also results in mass casualties. To avoid being thwarted by the enemy, these weapons would be designed to be extremely difficult to simply “turn off”, whereby humans could plausibly lose control of such a situation. This risk is one that is present even with narrow AI, but grows as AI levels of intelligence and autonomy increase.

2. The AI ​​is programmed to do something beneficial, but it develops a destructive method to achieve its goal: This can happen whenever we fail to fully align the AI's goals with our own, which is surprisingly difficult. If you ask an obedient smart car to get you to the airport as quickly as possible, it can get you chased by helicopters and covered in vomit, not doing what you wanted, but literally what you asked for. If a super-intelligent system is tasked with an ambitious geoengineering project, it can wreak havoc with our ecosystem as a side effect and see human attempts to stop it as a threat.

As these examples illustrate, the concern with advanced AI is not malevolence, but competence. A super smart AI is extremely good at achieving its goals, and if those goals don't align with ours, we have a problem. You're probably not an ant hater who tramples the ants out of malice, but if you're in charge of a green hydropower project and there's an anthill in the area to be flooded, too bad for the ants. One of the main goals of AI security research is to never put humanity in the position of these ants.

Stephen Hawking, Elon Musk, Steve Wozniak, Bill Gates, and many other big names in science and technology have recently expressed their concern in the media and through open letters about the risks posed by AI, followed by many leading AI researchers. Why is the subject suddenly in the priority?

The idea that the quest for strong AI would ultimately be successful has long been thought of as science fiction, centuries or longer. However, thanks to recent advances, many AI milestones that experts saw as decades away just five years ago have been reached, causing many experts to take seriously the possibility of superintelligence in our lives. While some experts still guess that human-level AI is centuries away, most AI research at the 2015 Puerto Rico Conference assumed this would happen before 2060. Since it can take decades to complete the security research required, it is wise to start it now.

Since Artificial Intelligence (AI) has the potential to become smarter than any human, we have no sure way to predict how it will behave. We cannot use past technological developments as a basis because we have never created anything that has the ability, consciously or unintentionally, to help us. The best example of what we could face might be our own evolution. People now control the planet, not because we are the strongest, fastest, or biggest, but because we are the smartest. If we're not smarter, are we safe to remain in control?

FLI's position is that our civilization will flourish as long as we win the race between the growing power of technology and the wisdom with which we manage it. In the case of AI technology, FLI's position is that the best way to win this race is not to stop the former but to accelerate the latter by supporting AI security research.


A captivating conversation is taking place about the future of artificial intelligence and what it should mean for humanity. There are fascinating controversies on which the world's leading experts disagree, such as: the future impact of AI on the job market; if/when human-level AI will be developed; whether it will lead to an intelligence explosion; and whether this is something we should welcome or fear. But there are also plenty of examples of boring pseudo-controversies caused by people talking to each other and being mean. To help us focus on interesting controversies and open questions – not misunderstandings – let's clear up some of the more common myths.

The first myth considers the timeline: how long will it be before machines surpass human-level intelligence? A common misconception is that we know the answer with great certainty.

A popular myth is that we know we will have superhuman AI in this century. In fact, history is full of technology excess. Where are these fusion power plants and flying cars that we were promised we would have until now? AI has also been repeatedly overrated in the past, even by some of the founders of the field. For example, John McCarthy (who coined the term “artificial intelligence”), Marvin Minsky, Nathaniel Rochester, and Claude Shannon wrote this overly optimistic prediction about what could be accomplished in two months with stone-age computers:

“We propose that a 2-month, 10-man study of artificial intelligence be conducted during the summer of 1956 at Dartmouth College […]. An attempt will be made to discover how to make machines use language, form abstractions and concepts, solve types of problems now reserved for humans, and improve themselves. We think that a significant advance can be made on one or more of these problems if a carefully selected group of scientists work on it together for a summer."

On the other hand, a popular counter-myth is that we know we won't get superhuman AI in this century. Researchers have made a wide range of estimates for how far we are from superhuman AI, but we certainly can't say with great confidence that the probability is zero this century, given the sad record of such techno-skeptical predictions.

For example, Ernest Rutherford, arguably the greatest nuclear physicist of his day, said in 1933 - less than 24 hours before Szilard's invention of the nuclear chain reaction - that nuclear energy was "moonglow". And Astronomer Royal Richard Woolley called interplanetary travel a “total sewer” in 1956. The most extreme form of this myth is that superhuman AI will never arrive because it is physically impossible. However, physicists know that a brain consists of quarks and electrons willing to act as a powerful computer and that there is no law of physics that prevents us from building even smarter quark bubbles.

There have been a number of polls asking AI researchers from here how many years ago they think we will have human-level AI with at least a 50% probability. All these polls come to the same conclusion: the world's leading experts disagree, so we just don't know. For example, in a survey of AI researchers at the 2015 Puerto Rico IA conference, the average (median) response was up to the year 2045, but some researchers thought hundreds of years or more.

There's also a related myth that people who care about AI think it's only a few years away. In fact, most people who are worried about superhuman AI think it's still at least decades old. But they argue that while we're not sure it won't happen this century, it's smart to start security research now to prepare for the eventuality. Many of the security issues associated with human-level AI are so difficult that they can take decades to resolve. So it's wise to start researching them now rather than the night before some Red Bull-drinking programmers decide to switch one.

Another common misconception is that the only people raising concerns about AI and advocating AI safety research are Luddites who don't know much about AI. When Stuart Russell, author of the AI ​​Standard Textbook, mentioned this during his conversation in Puerto Rico, the audience laughed out loud. A related misconception is that support for AI security research is extremely controversial. Indeed, to support a modest investment in AI security research, people need not be convinced that the risks are high, merely not negligible – just as a modest investment in home insurance is justified by a non-negligible probability that the house burns.

It could be that the media has made the AI ​​security debate seem more controversial than it actually is. After all, fear is sold, and articles that use out-of-context quotes to proclaim imminent doom can generate more clicks than nuanced and balanced. As a result, two people who only know the positions of each of the media quotes tend to think they disagree more than they actually do. For example, a ceiling skeptic who has only read about Bill Gates's position in a British tabloid might mistakenly think that Gates believes superintelligence is imminent. Likewise, someone in the pro-AI movement who doesn't know anything about Andrew Ng's position except that his quote about overpopulation on Mars might mistakenly think he doesn't care about the safety of AI, while in fact, it does. The crux is simply because Ng's timeline estimates are longer, he naturally tends to prioritize short-term over long-term AI challenges.

Many Artificial Intelligence (AI) researchers roll their eyes when they see this quote: “Stephen Hawking warns that the rise of robots could be disastrous for humanity.” And like many, they lost count of how many similar articles they saw. These articles are usually accompanied by an evil robot carrying a weapon, and they suggest that we should be concerned about robots rising up and killing us because they have become conscious and/or evil. On a lighter note, these articles are actually quite impressive because they succinctly summarize the scenario that AI researchers don't care about. This scenario combines up to three separate misconceptions: concern for conscience, evil, and robots.

When you drive, you have a subjective experience of colors, sounds, etc. But is a self-driving car a subjective experience? Do you feel anything in a self-driving car? While this mystery of consciousness is interesting in its own right, it is irrelevant to the risk of AI. If you are hit by a driverless car, it makes no difference to you whether you feel subjectively aware. Likewise, what will affect us humans is what superintelligent AI does, not what it subjectively feels.

Fear of machines becoming evil is another red herring. The real concern is not malice, but competence. A super-intelligent AI is, by definition, very good at achieving its goals, whatever that may be, so we need to ensure that its goals align with ours. Humans don't generally hate ants, but we are smarter than they are – so if we want to build a hydroelectric dam and there's an anthill there, it's pretty bad for the ants. The pro-AI movement wants to avoid putting humanity in the position of these ants.

The consciousness misconception is related to the myth that machines cannot have goals. Machines can, of course, have goals in the strict sense of exhibiting goal-oriented behavior: the behavior of a heat-seeking missile is economically explained as a goal to hit a target. If you feel threatened by a machine whose goals are out of line with yours, then it is precisely its goals in this strict sense that bother you, not if the machine is conscious and experiences a sense of purpose. If that heat-seeking missile chased you, you probably wouldn't exclaim, "I'm not worried, because machines can't have goals!"

I sympathize with Rodney Brooks and other robotics pioneers who feel unfairly demonized by creepy tabloids because some journalists seem obsessively focused on robots and adorn many of their articles with evil metal monsters with glowing red eyes. In fact, the pro-AI movement's primary concern is not with robots, but with intelligence itself: specifically, intelligence whose goals are out of line with our own. To get us into trouble, such scruffy superhuman intelligence doesn't need any robotic bodies, just an internet connection - it can allow us to outrun financial markets, invent human researchers, manipulate human leaders, and develop weapons we can't even understand. Even if construction robots were physically impossible,

The robot misconception is related to the myth that machines cannot control humans. Intelligence Makes Control Possible: Humans control tigers not because we are stronger, but because we are smarter. This means that if we give up our position as the smartest on our planet, it's possible that we might as well give up control.


Not wasting time on the aforementioned misconceptions allows us to focus on real and interesting controversies where even experts disagree. What kind of future do you want? Should we develop lethal autonomous weapons? What would you like to happen with the automation of work? What career advice would you give today's boys? Do you prefer new jobs that replace old ones, or an unemployed society where everyone enjoys a life of leisure and machine-produced wealth? Further down the road, would you like us to create a super-intelligent life and spread it throughout our cosmos? Will we control intelligent machines or will they control us? Smart machines will replace us, will they coexist with us or will they join with us? What will it mean to be human in the age of artificial intelligence? What do you want that to mean, and how can we make the future that way? Join the conversation!

Post a Comment

Previous Post Next Post