ChatGPT: Where do we go from here?

To say that ChatGPT has taken the world by storm would be an understatement. It has well and truly wedged its way into our collective psyche as the representation of what Artificial Intelligence systems are capable of. Given the fact that some of the first expert systems were conceived more than 50 years ago, the value and impact of AI today is not surprising. We have expected, and dare I say, even hoped, for AI systems to develop and come to the fore to help in the progress of humanity. With ChatGPT and its ease of use, access, and output quality, (and a lot of tolerance and understanding, I must say) the general public has now seen the true capability of AI. And based on the viral discussions on the technology, I would say they have embraced it, or at least tolerated the idea of AI in their lives. It is definitely a watershed moment!?

So what is it? GPT stands for Generative Pretrained Transformer (GPT), a highly developed (and trained) version of the transformer neural network model that came out of Google circa 2017 and was particularly well suited for natural language understanding. ChatGPT is a GPT that has been trained on incredible amounts of data and optimised for human conversation. Think of it as a Google search that distils the millions of search results into a paragraph that (most of the time) makes sense. But ChatGPT takes it much further, because it has “generative” in its name – it can make up an answer if one does not exist, based on the patterns it has already seen in the data.

For example, it can generate working Python code if you tell it to “write a program that plays tic-tac-toe" (based on the billions of lines of Python code that exists on the internet), or “write an essay on Christopher Columbus visiting America in 2015” (based on the various books and articles based on his life, as well as the description of the world circa 2015), or even write poetry in the style of Lord Byron on the status of ChatGPT (by learning the patterns found in the poetry and critical writings by and on Lord Byron). It may even have been able to clear Wharton’s MBA exam while it is at it! Or take ChatGPT’s cousin, Dall-e, which generated this image with the simple description of “a photo of an astronaut riding a horse” and can apply Dali, Van Gogh, or Picasso’s painting style to the generated image as well.?THIS is undeniably magic!?

But this is not new – large language models have been in the works and existed for a few years. Microsoft released Tay in 2016, an AI chatbot that was shut down within 16 hours because it “generated” inflammatory and offensive tweets. Google’s LaMDA has had a Google engineer go public with his belief that LaMDA is sentient! Meta/Facebook’s BlenderBOT, released to open source, also went through a tough phase. A much publicized response from BlenderBot was, "Since deleting Facebook, my life has been much better!” Doesn’t that bring a smile to your face? It also underscores the underlying sentiment and the love and hate relationship people have with social media. What OpenAI has done well with ChatGPT is made it available whilst acknowledging all the technology’s flaws and crowd sourced its refinement – the thumbs up or down that millions of human users are providing ChatGPT responses is allowing it to get better at a global scale.?

The growth of ChatGPT reminds me of the classic short story by Isaac Asimov called “The Last Question.” In the story, the all-powerful AI/computer called Multivac has been fed data for millennia and assisted humans in achieving extra-terrestrial existence. It can answer all questions, except the one that it has been working over for a very long time: “How was the universe created?” or “Can entropy be reversed?” Over time, it has accumulated all the data the universe has to offer and evolved into an entity that can exist in hyperspace. Until then, Analog Computer’s (an updated version of Multivac; AC) only response to this question asked countless times across millennia is “INSUFFICIENT DATA FOR A MEANINGFUL ANSWER!” Still, when the universe dies and AC cannot find the answer to reversing entropy, it decides to recalibrate its data in all possible combinations, in one final attempt to solve the problem. With the data of millennia, the solution to the reversal of entropy is indeed possible. Since, there are no humans to report it to, it decides to demonstrate the phenomenon and says, “Let there be light, and there was light…”?

To some, this may seem like a dystopian narrative but the idea is more to do with survival and how the creation of AI will enable human beings to continue to exist! I find that to be more hope-filled than despair inducing. ?

So where do we go from here? Open AI recently announced a tool that will allow us to distinguish between human and AI generated output! Such a logical step, isn’t it? With the Turing test in 1950 (yes, more than 70 years ago) came the realization that humanity would need such a test. Though the Turing test was passed by machines (with multiple controversies), these new generation models like GPT-3 will need even more sophisticated ways to determine “human” creation – and laws and regulations invariably follow such determinations by technology. This reflects what the world has done in the past with technologies like electricity, genetics, space, nuclear power, and social media. Humans have pushed the boundaries of nature (as they are thought to exist!) with these technologies and evolved when nature pushed back. We will continue to push the boundaries of AI until we hit or overcome them. Until then, I gratefully accept its wonders, let it make my life simpler, and let it make me feel more creative than I have ever felt!?

And what would be more appropriate a conclusion than asking this question to ChatGPT itself:?

Q: What is the future of humanity with ChatGPT??

ChatGPT: As an AI language model, I don't have personal opinions or predictions about the future of humanity. I can provide information and analysis based on available data and research, but the future of humanity will be determined by the actions and decisions of individuals and society as a whole. The impact of AI and chatbots like me will depend on how they are developed, used, and integrated into society. It's important for society to consider the ethical implications and potential consequences of AI technology as it continues to advance.?

?

How appropriate! Wouldn’t you agree? And hope that AI continues to feel like this in the future?!