Toward the end of his abbreviated life, Alan Turing was deeply invested in artificial intelligence–a major influence in computer science.
Alan Turing’s interest in artificial intelligence was so pronounced that he created a test for computers to gauge their level of intelligence, one we now refer to as The Turing Test. This serves as a measure to determine when a computer can be said to have human-like intelligence.
If someone were to converse with a computer and could not distinguish that it was not another human speaking, this would prove the computer has achieved human-level intelligence. The test is, therefore, quite simple.
ChatGPT is undoubtedly ace when it comes to passing the Turing test. This virtual assistant goes far beyond this—it’s even been used to help students pass equally difficult med school and law school exams and take care of homework for entire classes of students in the US.
ChatGPT can be confusing because people tend to impose human-like qualities onto its generative AI, which it doesn’t truly possess. If you’re unfamiliar with ChatGPT, this uncertainty may prove overwhelming.
Misunderstandings are common; however, because ChatGPT-like AI technology is becoming increasingly popular shortly, there is a risk that people may grant it capabilities that go beyond what it has – this could have dangerous consequences if utilized under incorrect assumptions.
Understanding Adversarial Generative AI And Its Benefits
A Generative Adversarial Network (GAN) is the foundational technology for ChatGPT, Stable Diffusion, and other Artificial Intelligences that can produce music, imagery, text, and more. This GAN platform is what fuels these AIs to create their masterpieces.
A GAN consists of two software models working together; a generator producing output and a classifier determining whether the data is legitimate. The generator and the classifier collaborate to achieve results.
The generator and classifier in a GAN work together word-by-word or pixel-by-pixel, reaching a consensus before shifting to the next element. In an iterative process, they battle it out to achieve the desired result.
Granularly, a GAN possesses the same creative abilities as humans and can recreate works with uncanny accuracy. Bit by bit, it can produce very similar outputs to what a human would create.
The generator is extensively trained with many input data to generate outputs. At the same time, the classifier utilizes its input to evaluate the effectiveness of what has been produced.
No one knows how human intelligence “creates” new things, as neurologists are still researching the topic. But, the results of this creative process can demonstrate a measure of human-level intelligence; after all, you can tell the tree by its fruits.
Does Passing The Turing Test Really Matter For CHATGPT?
Turing formulated his Test for Artificial Intelligence to have a conversation practically indistinguishable from talking to a rational human being. By requiring understanding on the machine’s part, as opposed to just responding to predetermined keywords, Turing intended to demonstrate whether or not machines had achieved true intelligence.
For Turing, being able to converse with a machine to the point where it was indistinguishable from a human was an indicator of intelligence, something no one could have imagined at the time. This included creating a vast repository of words with different ratings and weights dependent on the words that preceded them in sentences.
The math required to make such a calculation, which now would be an easy affair even for the most inexpensive smartphone in nanoseconds and was then thought too complicated even to be worth considering, was as intractable as counting the number of atoms in the universe — the same task taking all day on machines.
Sadly, Turing didn’t live to witness the current world where data science and chatbots are ubiquitous and modern computers rely on integrated circuits.
Chatbots, even before ChatGPT, were certainly making progress towards passing the Turing test as commonly perceived; people using them to communicate with banks can testify to that. However, Turing would not have identified such a chatbot as an artificial entity possessing a human-level mind.
A chatbot is a tool with a single purpose and is incapable of human-level intelligence. It will always operate within the boundaries set by its developers without taking any additional initiative.
GPTChat, like a parrot, can accurately recreate human language patterns; however, it is difficult to argue that understanding the real meanings behind words is also within its ability.
ChatGPT does not possess the same level of intelligence as a human – even though it passes the Turing test. The Turing test, however, cannot accurately assess artificial intelligence.
What Can CHATGPT Do? Understanding The Limitations Of AI Technology
US Senator from Connecticut Chris Murphy tweeted this week that ChatGPT had learned chemistry without any instruction from its creators. Unless additional information has not been released yet, it appears that the current version of ChatGPT does not have this capability.
If ChatGPT, or any other GAN, demonstrates independent initiative by seeking and learning on its own — unquestionably a marker of true human-level intelligence — we have undeniably crossed over into the post-human Artificial General Intelligence world.
No matter what a developer instructs a bot to do, even if the bot follows their command and carries out the requested task, that does not mean it is intelligent.t mean it is intelligent.
The distinction between obeying someone versus expressing one’s own free will can be rather hard to discern by an outsider. However, ChatGPT cannot recognize this difference unless it has been fed philosophical texts concerning free will.
Free will is something that, if it has never been exposed to Plato, Aristotle, or Nietzsche, an AI wouldn’t conceive of without being told so. It simply would not be aware of the concept’s existence.
What To Know About ChatGPT: Reasons To Worry & Benefits
ChatGPT poses a great risk due to its inability to comprehend what it generates. This can lead to defamation, plagiarism, and inaccurate information being presented as fact. To summarise, the dangers of ChatGPT are considerable and should not be underestimated.
ChatGPT and other generative AIs, such as Stable Diffusion, are raising an important question: can they be a substitute for human workers? After all, they appear to be more efficient, convenient, and cost-effective than manual labor.
Though discussing it is best left for later, the biggest peril of ChatGPT would be its social implications and not the possibility of constructing something like Skynet.
ChatGPT is a system composed of interconnected digital nodes and assigned values that generate coherent responses based on an input given. However, this is simple math programmed into a machine rather than thought. Intelligence lies in the human minds that created it, not to be confused with the complex output on display.
That said, developing language models like ChatGPT represents a significant milestone in artificial intelligence. These models have numerous practical applications and are constantly improving as researchers refine their algorithms and incorporate new data.
As we continue to explore the possibilities of AI, it is important to approach these technologies with a clear understanding of their limitations and potential. While ChatGPT may not be a true AI in the sense that many people envision, it is still an impressive feat of engineering and a powerful tool for processing and understanding human language.
Source: TechRadar
