"Common sense is not so common." – Voltaire

In a world where technology continues to develop at a rapid speed, the issue of 

'Is it possible to teach common sense to computers'  remains significant. Read till the end to find out the answer!

So, first - Common sense. We all think we've got it. So, what precisely is it? Can computers or artificial intelligence systems really learn it? 

What Is Common Sense, and How Do Humans Acquire It?

Common sense is the basic capacity to detect, comprehend, and assess things that most individuals are believed to possess. It is a collection of facts, knowledge, and general rules of thumb gleaned from our life experiences and observations. Common sense permits us to absorb and react to ordinary circumstances without overthinking them.

Humans begin to develop common sense early in development. As newborns, we begin to understand cause-and-effect linkages, such as how crying results in being fed or changed. Through repeated experiences, we build practical understanding about the world. For example, touching a hot stove causes burns. So we learn to avoid touching hot surfaces.

Our common sense evolves as we grow and are exposed to new individuals and circumstances. So, a youngster growing up in a small town develops fundamental common sense about life in that environment. An adult migrating to a huge metropolitan metropolis must modify their common sense to their new environment.

Common sense evolves when we acquire fresh experiences in our life.

Why Does Common Sense Challenge Computers?

There are several reasons why common sense is difficult to program.

For starters, individuals develop common sense throughout time as they interact with their surroundings. We experiment, observe what works and doesn't, and remember what we've learned. Computers lack those kinds of real-world experiences to rely on. They only understand what people expressly tell them.

Another difficulty is that common sense depends on context. If a computer just has pre-programmed rules, it cannot adapt to new situations in the same way that people can.

For example, suppose you trained a computer what to do when it began to rain outdoors. Seems simple, right? But what if, instead of rain, a sprinkler system is turned on? What if you're inside a grocery store and the pipes start pouring water from the ceiling? We'd quickly know how to manage such variances, but a machine would blindly obey its "when it rains outside, go inside" rule, which no longer makes sense.

How Can Computers Learn Common Sense?

After initial excitement in the 1970s and 1980s, researchers discovered how difficult teaching computers common sense would be. However, emerging techniques show potential in teaching AI computers fundamental common sense about the physical and social worlds.

One option is to manually create vast knowledge bases including facts and rules about how the world operates. Doug Lenat's Cyc project, which began in 1984, is one example of such an ambitious attempt.

Teaching Common Sense via Experience

Another interesting method is to create rich simulated environments in which AI agents may explore and acquire physics and intuitions via experience.

Researchers are constructing 3D virtual settings packed with ordinary goods that mirror the real world, such as the Allen Institute's digital house "AI2 THOR". Within these settings, AI robots may experiment with a variety of interactions to get an intuitive knowledge of ideas that humans take for granted.

For example, an AI bot can be given a virtual body and instructed to pick up, stack, and knock over bricks. The bot learns fundamental concepts like solidity, gravity, and physical dynamics by seeing the blocks fall and clash in a realistic manner. No rules are required—only experience.

The bot can also do behaviors such as dropping a glass object and seeing it break as it strikes the ground. Alternatively, it may test the characteristics of water by pouring liquids and watching how they flow and pool. These hands-on teachings base the AI's expertise on sensory experience rather than merely data patterns.

Progress Made, But More Work Needed on Common Sense AI

Some researchers claim that AI will never achieve human common sense until it develops brain structures and bodies similar to ours. On the other hand, computer brains are not constrained by human prejudices and mental shortcuts, thus they may hypothetically outperform humans! However, we probably don't need to worry about super-intelligent AI just yet.

In the short run, the greatest hope is AI that blends taught common sense and good old-fashioned programming. That way, stupid blunders like mistaking a turtle for a gun can hopefully be avoided.

We're not there yet, but common sense is no longer AI's dark matter--progress is being made! However, a good dose of human common sense will be required in using these technologies for some time.