This year’s IDC Directions event, held in Boston earlier this month, brought together dozens of experts and hundreds of attendees to discuss the latest over-arching technology trends in digital business. The event included numerous breakouts on cloud development, digital sovereignty and security trends, enterprise automation, and general updates into the evolving state of digital transformation.
Every major panel, however, including the event’s keynote, centered around the digital elephant in the room: artificial intelligence (AI). Since the release of ChatGPT, conversation has erupted around the use of AI, with some praising the arrival of a full-blown AI revolution, and others cautioning or outright dreading the arrival of what they see a full-blown human replacement for many industries.
So, what was IDC’s take? Before we get into that, let’s lay some groundwork: What exactly is ChatGPT, what does it do, and how has it ignited such a speculation frenzy surrounding AI?
To use the Encyclopedia Britannica definition, AI typically refers to “the ability of a digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings.” The “intelligent beings” aspect of the definition needs a little more clarification, it refers specifically to the mental processes we most commonly associate with humans. The ability to apply logical reasoning, for instance, or generalize a situation, or learn from past mistakes. These qualities (though increasingly observed in the animal kingdom) are still considered human, and having a computer replicate them marks the divide between automation and AI.
With this benchmark, what exactly is ChatGPT and how is it different from the chatbots that have come before? To be clear, we’re going to stay more surface level here. There is a lot to the depth and complexity of ChatGPT, and to explain it all would take another article entirely. For example, we will not be focusing on its ability to write music or create stories (although every path of creation functions the same), but rather its more business-centric capabilities.
To sum up, older business AI chatbots were predominately designed to help businesses assist their customers by connecting them to appropriate existing resources. If the user had a question, then the chatbot would act as a sort of directory assistant, retrieving the appropriate material based off an existing set of anticipated, preset responses. Right here on this blog post, you’ll probably see our chatbot virtual assistant functioning exactly in this manner, taking you through a list of the most commonly asked questions and giving relevant responses.
ChatGPT goes far beyond this functionality. The GPT in ChatGPT stands for Generative Pre-training Transformer, which sounds intimidating, but simply means it is part of an AI subset that can pull from an incredible amount of data, including the internet, books, and academic articles (all published in 2021 or prior). From this massive dataset, it has been “trained” to understand the patterns and predictability of human language. ChatGPT does not read the way people do, instead it uses a large language model (LLM) to calculate the way language works.
It was trained/improved through reinforcement learning from human feedback (RLHF), wherein OpenAI hired a team of human specialists to provide a mix of supervised learning and reinforcement learning, thus refining and rewarding the AI responses to provide better accuracy. Even users today can upvote or downvote ChatGPT’s responses, which is another form of feedback and learning for the program.
Despite what you may have read, however, ChatGPT has very definite limitations. Everything in the program’s dataset comes from 2021 or before, so it has no innate knowledge of recent events. Likewise, as one of its data sources is simply “the internet,” this means that not all of ChatGPT’s source knowledge is accurate. Lastly, it is a program, not a person – it does not think. We are still a long way away from truly sentient AI capable of turning over a problem and applying creative problem-solving. This means many logic puzzles and riddles are not properly processed.
Despite these limitations, there is no denying how powerful a tool ChatGPT is, and we’re just at the beginning. AI, with its ability to process large datasets in seconds, is equipped to perform tasks that no human could reasonably do. In the medical field, for instance, imagine a similar program to ChatGPT that, instead of processing language, looks at x-rays or CT scans to provide medical diagnosis. It can be an incredible tool, and that was the message of the IDC Directions’ keynote from David Autor, Ford Professor of Economics at MIT. , took to the stage to discuss AI in the context of work.
Rather than labeling it as a job eradicator, Autor expressed confidence that humans will remain a vast majority of the workforce, citing historical trends. Those curious to see a similar presentation can check out this TED Talk Autor recorded six years ago. The focus may be automation instead of AI, but many of the trends and conclusions remain the same:
Of course, AI is arguably the most powerful technological tool we’ve developed. Autor acknowledged this and told the audience that, essentially, the ball was in our court. AI is not sentient. It does not think about what to do next. It must be directed; it has to be programmed with specific goals in mind.
Autor showcased two examples to illustrate this dramatic potential. The positive is happening in Japan, where AI and robotics are being applied to advancements in elderly care. The negative, he said, is happening in China, where AI is being used to implement the most comprehensive surveillance state that has ever existed. The potential of AI is vast and, looking back in 10 or 20 years, ChatGPT will be seen as only the beginning.
It is fitting that this event was IDC Directions, as “direction” may have been its most important theme. Where AI goes next depends on ethical companies and governments striving to do good, as well as humans looking to give ourselves a new, powerful set of tools to accomplish and create far more than we ever could before.
Colin McMahon is a senior market research analyst working with PTC’s Corporate Marketing team, helping to provide actionable insights, challenging perspectives, and thought leadership on trends, technologies, and markets. Colin has been working professionally as a research analyst for many years, and he enjoys examining and evaluating just how large the overall impact of digital transformation technologies will be. He has a passion for augmented reality and virtual reality initiatives and believes that understanding the connected ecosystem of people and technology is key to a company fully realizing its potential in the 21st century.