The future of GPT-3 and AI

- While GPT-3 today is remarkable and valuable on many levels, in my view, one of its most important qualities is how it may inform us about what is ahead for AI. Let's first look at what GPT-3 may mean to customer service. Many of us have likely already experienced the increasing use and value of chat bots. A chat bot is online software that enables a human to converse with a computer. Typically, we see it in customer service. You might visit a website, and a chat bot will pop up asking how it may help you. To date, these have supported relatively simple exchanges. Asking for product prices and availability or phone numbers is common. The chat bot does a simple search and returns the answer. However, complicated exchanges have been elusive so far. Entering a complex question usually results in a standard answer, suggesting it doesn't understand the request, or it just refers the users to a human customer support. GPT-3 is changing this. First, it provides the possibility of a much better customer experience by broadening the ability for replies with details and accurate answers. However, as more questions are answered and more insights are derived from customers who ask the questions, the chat bot becomes smarter through transfer learning. This is a machine learning method that uses previous insights as starting points for new requests. It's like when a human learns to paint with watercolors. When they later use oil paint, they don't have to relearn how to hold the paintbrush. As AI develops, the chat bot conversations will seem more human, because the further the conversation progresses, the richer the dialog. Unlike early versions of AI, where each question and answer exists ephemerally, the software will retain and build an attained knowledge. Next, let's briefly look at what GPT-3 and its future may mean to our work. It's reasonable to acknowledge that skills and experiences limit many people from participating in certain work. That's why we go to school and pay for expensive education. For example, it's not possible to simply sit down in front of a computer without specific skills and write a complex software program. However, what happens when AI acts as a mediator that converts plain English into code that normally requires deep skills? We get something called low coding, or with GPT-3 and its successors, no coding. Suddenly, high demand highly-skilled work has been devalued, and simultaneously the opportunities for innovation explode. It's a double-edged sword. Other work that will likely be automated by GPT-3 and future AI ranges from the explanation of legal documents, translation services, medical answers, solving math problems, content development, such as news articles and blog posts, and even website creation. Sure, today we're still a little distanced from any significant job impact from GPT-3. But can it be too long as new versions and improvements emerge? The leaps in performance demonstrated by GPT-3 establish a new research baseline. We can assume that whatever comes next will be more powerful. Many have pointed out clear GPT-3 limitations and errors, but as it matures, we can expect these to fade away. AI today still can't pass the turing test that's still the respected bar for determining where developments stand. It's only a matter of time before AI aces this test. Then we'll see AI augmenting and replacing any number of interactions that would typically be staffed by a human. Why not? It can work 24/7 without pay, never get tired, and it can almost infinitely scale. Finally, GPT-3 considerably raises the bar on our expectations of AI. But let's not get ahead of ourselves. GPT-3 doesn't get as much closer to sentience, emotions, human-like reasoning, and consciousness. That said, humanity appears determined to reach those goals.