It’s been a year since OpenAI quietly rolled out ChatGPT as a “research preview,” creating a chatbot based on a large language model (LLM). LLMs are a specific kind of transformer neural network technology, first introduced in a 2017 paper by Google. ChatGPT provided an easy-to-use interface for the underlying LLM, GPT-3.5, and quickly became the fastest-growing consumer technology ever, gaining over a million users within just five days of its launch. Today, there are hundreds of millions of users of ChatGPT, alongside numerous similar bots from various companies, including Amazon Q, a business-focused chatbot.
These AI technologies may fundamentally change creative and knowledge-based work. A study by MIT last summer looked at tasks like writing cover letters, delicate emails, and cost-benefit analyses, and found that using ChatGPT reduced the time workers needed to complete these tasks by 40%, while the quality of the output, as rated by independent evaluators, increased by 18%.
People often compare AI to groundbreaking discoveries like electricity and fire because it has the potential to significantly alter many facets of our lives, from work and communication to solving complex problems, much like how electricity revolutionized power and industry, and fire transformed early human society.
Over the past year, the advancements in AI have been rapid, almost like moving from the Stone Age to the Space Age. Consulting firm McKinsey estimates that generative AI could add over $4 trillion annually to the global economy. As a result, tech giants like Microsoft and Google are aggressively exploring this market.
There have been ongoing debates about the impact and safety of AI since ChatGPT’s debut. These debates, spanning from the U.S. Congress to historic locations like Bletchley Park, generally fall into two camps: AI “accelerationists,” who push for rapid AI advancement to leverage its potential benefits, and AI “doomers,” who urge a cautious approach to mitigate potential risks.
This debate has led to significant steps in AI regulation. While the European Union is working on the AI Act, the U.S. has advanced with an Executive Order titled “Safe, Secure, and Trustworthy Artificial Intelligence,” aiming to strike a balance between rapid development and strict oversight.
Countries around the world are actively developing AI strategies in response to the rise of LLMs. Recently, Russian President Vladimir Putin announced new plans for AI development to counter Western dominance in the field, even though the U.S., China, the U.K., and other nations are already well ahead.
The past year has been a whirlwind in AI. A recent notable event was the firing and quick rehiring of OpenAI’s CEO, Sam Altman, after a significant pushback from investors and employees. Now, a new project named Q (pronounced “Q-star”) has surfaced as the next big topic. This secretive project was kept under wraps until just before Altman’s temporary departure. There was a letter from researchers warning the OpenAI board that Q could pose a threat to humanity.
There’s speculation about what Q might be. Some suggest it’s an innovative neuro-symbolic architecture, while others think it’s an advanced combination of existing LLM techniques. A true neuro-symbolic architecture could enable AI to learn with less data and to explain its decisions more clearly, a goal many companies and academic institutions are striving for, including IBM, which sees this as a pathway to achieving artificial general intelligence (AGI). AGI is envisioned as AI that can process information at or above human levels.
Although such advancements might not yet be here, if Q progresses to market, it will be another step towards AGI. Nvidia’s CEO believes AGI could be achieved within five years, while Microsoft’s President remains skeptical, suggesting it may take many more years, if not decades.
In the past year, breakthroughs like ChatGPT and potential technologies like Q have generated a mix of excitement, worry, regulation, competition, and speculation. These rapid advancements in AI are not just technological milestones but also reflect our ongoing quest for greater knowledge and control over our creations. The next year promises to be just as thrilling and uncertain, depending on how we manage and direct this energy and innovation.