Let's Give Common Sense To Artificial Intelligence

Introduction

Even with the dumbest humans, there is an aspect of our intelligence that is actually preserved for everyone as humans (or maybe almost everyone) - and that's the "common sense". This has to do with the perception of facts and the ability to pass judgment based on the perceived facts. Even though it has been said times and again that common sense may not be as common as it sounds, but to a certain degree, it is actually common, and also to a different level, every human is endowed with it. Example; when you give food to a kid, what stops them from putting it into their nose or eyes? That's commons sense. Now come to think of this: The idea behind the advent of Artificial Intelligence (AI) is to create a system that would enable the human cognitive abilities to be replicated in machines. But the area of common sense has been kept in the grey zone; you'll find out why soon, and the plans to get some common sense put into AI. Welcome again to the future.

_{[Image Source: Maxpixel. CC0 Licensed]}

The question I always ask myself each time the subject of AI is raised is: Have we truly achieved an absolute intelligent machine? The truth is; some of the Artificial Intelligence we have around - "the reactive machines, and the limited memory AI" would only relate to a pre-defined database with the help of some sets of algorithms to perform an action or draw an inference, but would you call that common sense? Of course not. As it were, the major advances in the field of AI has been to make them perform complex tasks with precision; even way beyond the human ability; but what of simple tasks that your kid in the nursery school can perform? Take this instance: If you ask the question "how many months in a year have 27days?" - A reactive machine would try to relate to the database to find a possible solution, and since no month is fed into the database to have 27days, it would give off an error message. But common sense should tell you that every month has 27days, lol.

That's it with common sense. But with the advent of Machine Learning, machines have now been incorporated with the ability to acquire new sets of information over time; through relationships and experiences. And that has even gone further into Machine Teaching - in which the knowledge acquired by these machines can be passed on to another machine. This is actually what has been leveraged upon to make the idea of "teaching" common sense to Artificial Intelligence.

Hey Ai!! Have some sense

How do you know that "fire" burns? I'm sure most of us figured this out ourselves. I remember my friend's little son. Every time, the mum would be like; "Hey Jay, don't go near the burner, it's gonna burn your hand", but Jay would only prefer to play around the burner, until one day, his hand went on the burner and he got some nasty burns... and ever since then, he quit playing around the burner. That's exactly how machine learning works - relating with past experiences to solve future problems. Not just that; the little kid; Jay (our case study, lol) went on to tell his other siblings of his experience; and they also learnt from his experience not to play around the burner - that's how Machine Teaching works; passing on acquired info.

_{[Image Homemade by me: @samminator]}

Up till very recently, every attempt to teach these machines some common sense has been seen as "mission impossible" - partly because there is no standardized scale for measuring common sense. But a headway has been made in this by a team of researchers from the Allen Institute for Artificial Intelligence, and they came up with the perfect plan - Project Common-sense; you'll see wassup with this project soon.

Here's a little spoiler's alert for y'all before I continue: I'm sure you may have known or have heard of the badass AI that plays the game of GO; the AlphaGO. Brilliant AI; you may say; at least, it was seen to have put up a strong opposition to the human GO players, and even defeated the best of the best. But humorously; on analysis; it was discovered that this AI wasn't even aware that GO is a subcategory of a board game _[ref]. How common is common sense? Clearly, the AlphaGO lacked the basic human-level of common sense. That's exactly what "Project Common-sense" is set to do - to incorporate the human-level of common sense in machines.

But how da heck do they hope to achieve this? Remember what I told you about Machine Learning; this would come in handy here. But first of all; it is worthy to note that humans (which is what the AI is trying to mimic) have the ability to mix and switch from one thought to another, and this particular ability has given a unique identity to our reasoning and thinking pattern. So to effectively simulate these human thought patterns, there would be a need to sample information from large dataset of people; and these infos would be collated into a single repository; of which analysis would be made on these data to create a pattern to relate and communicate these data to machines. Also, in addition to info from people, info can also be gotten from textual files, even from multimedia. Just like it is depicted in the image below.

_{[Image Homemade by me @samminator]}

From the image above, the repository hosts the collection of data. But it doesn't stop there; some specific algorithms would be used to transfer these patterns from the repository to the host machines. But in the real sense of it; this is just like the transference of common sense from one entity (humans) to another (machines). And that would form the bedrock upon which the Artificial Super Intelligence (ASI) would be built. And of course; this would be the true definition of AI, that would be modeled to simulate every of the human's domain of cognition to the minutest details. Maybe we'll take a quick peek into what ASI is.

Have you heard of the phrase that "AI would outsmart humans"? Now, that's the ASI that is being referred to here. Though currently, it only exists in theory, but it doesn't make it non-attainable. Have you ever imagined AI becoming self-aware, conscious, and even emotional? Not just that, but even exhibiting the human-level of common sense. I'm sure most of you may have seen the movie "Ex machina" (definitely not @lemouth, because he once said he's not a movie person, lol), a particular character there; "Eva" is a perfect description of the technology of ASI, with all the human-like perception of judgment.

But the question is; why do we even need such advances? Thinking of it this way; even with our natural complexities as biological beings, there are still inherent limitations in our cognitive abilities. Also, this era has been witnessing outburst of daily data-rollout in the exabytes range. And no doubt, these data are way beyond the computational capacity of our brains to handle. But imagine if a system (ASI) is designed, which can provide assistance with this. Maybe these stats might interest you: Some decades ago, some of the most advanced computers in the world would only perform just a handful of calculations in a second. But in this era, we've been greeted with Cray super computers; like the beast that was unleashed recently - the "Tianhe-2", which has the ability to crack a whopping 33quadrillion calcs in just one second _[ref] - does this give you an idea of what to expect with the ASI.

_{[Image Source: Pixabay. CC0 Licensed]}

Imagine when this insane processing power merges with the ASI, and also with the incorporation of common sense, this might alter the fate of humanity greatly. This would also find practical applications in almost every aspect of our lives. Imagine that a period is approaching in which we would co-exist together with machines freely (and the process has already been heralded), so we would need them to think like us, so as to make for effective symbiotic relationship - and that's the purpose of incorporating common sense in AI. Just like I've always said; AI is the future, but the future is already with us.

Conclusion

Common sense has been looked upon as a quality that is precisely reserved for us as humans, but with the advances here and there, we could be sharing this quality with machines. Isn't it awesome that the features of the organic humans are now being replicated in machines. What can't technology do? Y'all should just brace up for what's coming ahead.

Thanks for reading

References for further reading:

• Computational Intelligence.pdf
• Ascribing mental abilities to machines.pdf
• Allenai/common-sense
• consciousness in AI

gif by @foundation