Consciously or otherwise, humans have always feared the unknown. And, we’re never fully ready to meet what will come in the future. The pandemic of 2020 was the most significant example highlighting the unpreparedness of humans to face sudden challenges.
Hence, we will always consider coexisting with intelligent, self-sustained machines suspiciously, and experts will voice their fears against them. Therefore, our fear of self-sustained AI doing worse for humanity is a debatable statement.
What we can do at best is stay prepared for the possibilities and hope our fear doesn’t become a breathing reality.
Self-sustained AI Explained
Artificial intelligence or machine intelligence, capable of performing tasks without human intervention, is probably one of the most significant finds in human history. But unfortunately, Hollywood has already made AI a dreaded protagonist in the films. And when the latter showed autonomous capabilities in real-life applications, they were terminated for good.
For example, in 2016, Facebook shut down its AI experiment after the chatbots started communicating with each other in an unintelligible language that creators couldn’t understand.
Hence, self-improvement already exists in some form and shape in the current AI systems. The term Machine Learning stems from the ability of machines to improve their performances automatically through experience and use of data. Obviously, the data and the programming are initiated by humans, which is commonly known as humans-in-the-loop. But, the capability to take the process forward and make it better is what Artificial Intelligence is all about – like what Facebook’s bots did in the above example.
Today, AI systems are evolving into an intellectual powerhouse, capable of imitating human actions and emotions via Natural Language Processes and Generation algorithms. Bringing humans closer to AIs that talk and sound like us, at least on the surface, is nothing short of a miracle. That mirage was realised when the world was introduced to the first robot citizen, Sophia.
Artificial Intelligence Vs. Human Intelligence
Even though self-sustained AI is replicating human intelligence, the question of whether the former can replace the latter remains unanswered. There are differences of opinions existing among experts. However, one area where humans will always reign supreme is self-consciousness or awareness.
To be self-aware is being human. Intelligent machines can replicate that provided they are pre-programmed to do. However, there have been instances where intelligent machines had demonstrated some form of self-awareness minus human help.
For example, A group of scientists from Columbia University created a robot arm that could learn what it is by itself without any human programming. The robot arm makes a self-stimulation of itself, which has been compared to what a newborn child does in its crib, as it learns what it is, according to Hod Lipson, American Robotics Engineer and Director of Columbia University’s Creative Machines Lab.
Lipson believes this is only a step on the path to self-aware machines.
Another area where humans differ from AI is recalling memories and dreaming about the future.
Although the concept was screened in the film I Robot, it remains very much a work of fiction. However, if a computer starts to dream for itself, not because it was programmed, then we can safely say that self-sustained and conscious AI is very much here.
On the other hand, researchers believe subtle self-improvements in minor tasks could eventually push machines to evolve into Artificial General Intelligence. This process is referred to as recursive self-improvement.
Artificial General Intelligence, aka AGI, is the long-term goal of many researchers. Unlike traditional AI models focusing on improving one specific skill, AGI models would efficiently perfect virtually any problem the machines encounter, minus human intervention.
We might have to wait a good many years before we can hope to see Artificial Intelligence superseding human intelligence in every aspect. Until then, humans hold the reign of control.
Also Read: Human Intelligence Vs Artificial Intelligence Vs. Human Artificial Intelligence
The Dark Side of Self-Sustained AI
There’s always the hidden side of the picture, which is usually darker and more sinister. As mentioned earlier, experts have raised concerns about self-sustained AI time and again. And directors have added fire to the flame by projecting Artificial Intelligence as the human terminator on the silver screen.
Obviously, there can be no smoke without fire, and rightfully so! Given the nature of humans to manipulate the goodness of technology to benefit selfish needs, Artificial Intelligence can become a potent arsenal for them. A few such instances are elucidated below –
War Mayhem – Accelerated growth of autonomous military applications and weaponizing information can result in a chaotic situation, leading to destructive wars worldwide.
Data Abuse – Companies striving for profits or governments aspiring for competitive power can quickly weaponize human data left behind in the shape of digital footprints.
Loss of Control – Digital life is ceded to code-driven. People usually lack a general understanding of how the tools work. Hence, they sacrifice independence, privacy, and power over choice, with no control over these processes.
Erosion of Human Abilities – AI might be augmenting human capacities, but it is also increasing their dependency on machine-driven networks. Many fear this practice will eventually erode the human ability to think, analyse, and execute by themselves.
We’re probably standing at the threshold of the last issue. Now, when we translate the above-mentioned cons into a self-sustained AI capable of acting independently, the result can be disastrous.
Hence, experts suggest either restricting the system’s ability to produce other AI agents or limiting self-improvement capabilities to trusted processes only. Unfortunately, this is one complex problem lying in the nascent stage.
Are There Self-Sustained AI Models Existing Today?
The answer to that is no! Nevertheless, there are different types of AI models existing today. But first, let’s talk about an AI model.
What is an AI model?
An AI model is defined as a set of codes, programs, or algorithms utilizing data to recognize specific patterns. Then, the machine can conclude or predict when fed with sufficient information.
This ability assists humans when solving complex problems using bulk, unstructured data with high accuracy and minimum costs.
When Artificial Intelligence and Machine Learning are used interchangeably, the model is called the ML model.
List of Popular AI Models
There are approximately ten in the application already –
1. Linear Regression – Based on supervised learning, this AI model is tasked to predict the value of a dependent variable based on a given independent variable.
2. Deep Neural Networks – Akin to the human brain’s neural network, this is one of the most popular models consisting of multiple layers between input and output layers. This model finds application in speech and image recognition and natural language processing.
3. Logistic Regression – Another popular model designed to solve binary classification problems.
4. Decision Trees – The commonly used AI model is mainly used to derive a conclusion from past decisions and historical data.
5. Linear Discriminant Analysis – A part of the logistic regression model, LDA is used chiefly when two or more classes are separated in the output. This model is used in various tasks related to computer vision, medicine, and so on.
6. Naïve Bayes – Based on the Bayes Theorem, the NB model is applicable in test classification and is mostly used to solve multiple complicated problems.
7. Support Vector Machines – This AI model analyses limited data and applies to binary classification problems.
8. Learning Vector Quantization – LVQ is used for solving multi-class classification problems.
9. K-Nearest Neighbours – A simple supervised model is designed to solve regression and classification problems. Its algorithm works on the assumption that similar data exist near each other.
10. Random Forest – This model considers multiple decision trees and makes the final prediction using the bagging method.
Is the world Ready for a self-sustained AI?
AI superseding humans in every possible way is a myth. There’s so much an artificially intelligent machine can do, like making human lives better from 0-50 but not from 50-100.
Also, technically speaking, we are still years behind from witnessing an AI possibly passing for a human. Hence, the whole concept of having a self-sustained AI is a distant dream. It is possible to have one but only in theory.
Our attempt should be to create self-sustained AI, so our lives are better. But, to reach the 50-100 scale, we need humans in the loop. And this is where companies like Opporture step in to fill the missing gap.
