The Wonders of Generative AI in 2023: A Fascinating Revelation with a Cost

With all the astonishing breakthroughs and cultural impact of artificial intelligence (AI) in 2023, it would seem reasonable to crown it as “The Year of AI” – except that it’s been done before.

In fact, this academic journal has given that title to 1980, declaring it as the year of AI 43 years ago. AI has been around for a very long time. Back in the day, I wrote an article about AI for Computer Design Magazine titled “Artificial Intelligence as a Systems Component”. By 1988, two AI-based products for the Mac were introduced. Even then, AI was already over 30 years old. We can trace some of the earliest AI activities back to Professor John McCarthy of Stanford, MIT, and Dartmouth, who founded the Stanford AI Lab in 1955. Fast-forward to the present, and AI is celebrating at least 68 years of existence. The speculative fiction side of AI goes back even earlier, as Isaac Asimov started to explore AI ethics in 1940.

This foundational work aside, it’s hard to argue against calling 2023 the Year of AI. It’s been an eventful year. What changed? AI has been in use for a very long time, finding its way into expert systems, diagnostic tools, video games, navigation systems, and many other applications for decades. However, in 2023, AI has really come into its own as true generative AI. In previous years, a lot has been claimed as “The Year of AI,” but there’s no denying that 2023 is the “Year of Generative AI”.

So, what truly distinguishes this year is the profound advancements in training AIs. Previously, most AI training had been supervised, providing the AIs with specific information designed by their creators to form their knowledge base. This limited the scope of what the AI was capable of. In the current era of large language models (LLMs), pre-training is now unsupervised. With OpenAI leading the way, AIs are being trained with a broader range of information, drawing from virtually all available digital content, including the entire internet. This approach has enabled AIs to produce incredibly diverse material. Furthermore, significant strides in processor performance and storage have also played a role in making large language models possible.

To illustrate this point, we can compare the products developed years ago with today’s ChatGPT. House Plant Clinic was a successful expert system trained in plant-based knowledge and created with painstaking effort. In contrast, ChatGPT can confidently discuss almost any subject, but it relies on vast pools of unverified data, leading to less accurate information.

Generative AI is truly remarkable. Throughout the year, I have used it for various purposes and with great success. However, it is not without its challenges. The omnifarious knowledge on which it feeds has caused issues with accuracy and brought to light concerns around bias and discrimination.

Looking ahead, the evolution of this technology may lead to job displacement, starting with entry-level positions being replaced by AI services. As AI continues to advance, this trend may push experienced workers out of the market, compounding the issue further.

In conclusion, the events of 2023 have indeed highlighted the extraordinary potential of generative AI while also bringing to the forefront concerns about its influence on the workforce. This “Year of AI” has given us much to consider.