With all the regular updates about AI and various businesses trying to launch and relaunch AI vertices, it is high time we must address the question, “Can AI be smarter than humans?” In this regard, you will find differentiated answers united by a common belief that there are chances for AI to outsmart humans in the near future. However, no one is that sure about predicting an exact date. Or how is everything going to work? The write-up will help you with a thorough guide.
But before you look into “How can AI be smarter than humans in the near future,” let’s help you with a famous theory by Gordon Moore for a steady buildup on machines having the upper hand over their creators.
Understanding Moore’s Law
Gordon E. Moore (co-founder of Intel) predicted that the number of transistors packed in a single space would double every twenty-four months (two years). His law implied that computers and machines run by computers would eventually become faster, smaller, cheaper, and accessible to all with the upcoming generation.
And now, after 60 years of his prediction, we all can see the real-time effect of Moore’s law on modern gadgets, especially computer systems today. Microscopic microchips store all the data, fitting in information for an entire company or state. Also, the price of high-power computers has been coming down.
Can AI Be Smarter Than Humans? A Comprehensive Outlook
With ChatGPT and GPT 4 in the scene, there are speculations about AI being able to outsmart humans in the near future. However, regarding subjective feelings, experiences, and empathy, AI still lacks the basics dominating the human world. Its functions are merely based on algorithms that respond to situational data and several conditionings. For example, in GPT 4, you need to make a command specifying that the AI tool can use an external prompt, and only then will it start working accordingly. AI lacks on the part of decision-making on its own.
However, certain learning techniques, which, when organized chronologically and the AI made a part of the process, might improve the cognitive behavior of the system. Thus paving the way and making some solid revelations to the question, “Can AI be smarter than humans?”
Reinforcement Learning
Beginning with the first technique – Reinforcement learning, is a machine learning method based on rewarding the desired/positive behaviors and punishing the negative ones. The reinforcement learning agent learns to take the appropriate actions through trial and error.
So far, developers have already been using the reinforcement learning technique to train and twitch AI’s natural level of understanding. In the process, they have developed a loop based on “penalties” and “rewards.” The scientists devised a list of successful and unsuccessful outcomes, which the AI uses to adjust itself.
And already, as a result of such rigorous development and training program Google DeeMind’s AI AlphaGo has been able to defeat a professional human Go Player and score an otherwise quite complex goal.
Conditional Reasoning
Moving on, deciphering the appropriate answer to the question, “Can AI be smarter than humans?” It is just the tip of the iceberg that we are looking at. And on receiving proper training, things can be better than they are today. AIs made part of conditional reasoning programs can work better.
In “Conditional Reasoning,” the reasoner draws a conclusion based on the surrounding conditions and decides their further actions. For example, a reasoner will have these thoughts in their brain, “if it is Monday, then I will be attending the music class today.” As a reasoner, you can see yourself drawing answers based on the information available.
Answering – Can AI Be Smarter Than Humans?
As we all know, it took continuous training for the AI AlphaGo to learn about the basics of the football game and beat a professional human Go Player. The AI played millions of games and accumulated thousands of years of knowledge.
Now compare it with a human who has never played football and had never even thought of researching anything about the game. It would have taken a lot of time for someone with an average IQ to learn accurate dribbling skills and prepare for a professional-level match. However, the AI only took a few days to train, learn, match up, and beat a professional.
The AIs we are dealing with now are at their intermediate levels, with so much information yet to be fed. And after a few years, probably with GPT launching its 10th or 12th version, things can be very different. By then, AI will have all the information it can use to condition ‘reasoning’ and come up with answers that may lead to a smarter outcome, possibly better than the human brain could ever imagine.
