ChatGPT, False Idol

ChatGPT is revolutionary in its own way. But it has lagged behind in key features that would have made it magnitudes more powerful.

Many have marked the emergence of ChatGPT as the harbinger of the new AI revolution. It has reached 1 million users quickly than any other app, and raised a lot of questions about the role of AI in our economy and our society as a whole. It is because of ChatGPT that we now have these increasingly widespread predictions about AI taking over everything in the near future.

Indeed, the large language model (LLM) that ChatGPT pioneered is a game-changer: it manages to make itself look just like a genuine human with a great level of expertise across a wide range of topics, and has the ability to augment, if not replace, many types of jobs.

But many issues have arisen with the chatbot since its debut, and these issues make ChatGPT appear less complete than the competitors that have emerged in the months since its launch.

Much has been made of ChatGPT’s more-than-occasional parroting of incorrect information as fact. This is most likely because ChatGPT did not have connection to the Internet and has a knowledge cutoff of September 2021. More recently, a plug-in was unveiled to allow internet access, but the rollout so far has been rather rocky.

To me, the ability to connect to the Internet should be a gold standard for an AI chatbot. Big data is literally the fuel on which AI runs because it provides training for AI so it can respond to any kind of situation. And there is no bigger source of big data than the Internet. 

ChatGPT right now is basically a glorified data bank that was not able to take in new data by itself unless OpenAI trained it. While ChatGPT has received much acclamation since it’s release, and many of it is well-deserved, it simply would not be as competent in the future in its current form, because the Internet grows exponentially. Without Internet access, OpenAI would have to train ChatGPT itself on the ever-increasing amount of data that is in existence, a task that may only grow harder with the Internet’s exponential growth.

ChatGPT may be the world’s first online LLM-powered chatbot. But the type of chatbot that will truly own the future of generative AI is something like Google Bard and Bing Chat (which admittedly integrates ChatGPT within Bing), which can learn as the Internet expands because of its ability to connect to the Internet. This ability will ensure that its capabilities are relevant far into the future, and ChatGPT has been a follower in this regard. Therefore I think Bard should be the rightful symbol of the AI revolution instead of ChatGPT.

Leave a comment