It has been a year since OpenAI quietly released ChatGPT, a large language model (LLM), as a "research preview". LLM is a specific implementation of transformer neural networks, which was first introduced in a 2017 paper by Google.

ChatGPT provides a user-friendly interface for the underlying LLM GPT-3.5 and has become the fastest-growing consumer technology in history, with over one million users within the first five days of its launch. Today, the user base of ChatGPT has reached hundreds of millions. Moreover, there are many similar chatbots built on different LLMs from multiple companies. The latest one is Amazon's Q, an enterprise chatbot.

These technologies have the potential to completely disrupt creative and knowledge work as we know it. For example, a summer research conducted at MIT focused on tasks such as writing cover letters, sensitive emails, and cost-benefit analysis. The research showed that using ChatGPT "reduced the time it took for workers to complete tasks by 40% and improved the quality of output measured by independent evaluators by 18%".

People compare this technology to electricity and fire because, like these foundational discoveries, artificial intelligence is a transformative technology with the potential to fundamentally change almost every aspect of our lives, from how we work and communicate to how we solve complex problems.

Heading Towards the Future

Consulting giant McKinsey estimates that generative AI will add over $4 trillion to the global economy annually. Therefore, tech giants including Microsoft and Google are actively driving this market.

Since the emergence of ChatGPT, debates surrounding this technology and its safety have been ongoing. These debates, ranging from the US Congress to Bletchley Park, the site of the UK's secret code-breaking activities during World War II, can be broadly categorized into two viewpoints: AI "accelerationists" and "doomsayers".

Accelerationists advocate for the rapid development of artificial intelligence technology, emphasizing its enormous potential benefits. On the other hand, doomsayers lean towards a cautious approach, highlighting the potential risks associated with unrestricted AI development.

This has prompted substantial action towards regulating artificial intelligence for the first time. While the European Union's AI law, which has been in development for several years, may become a reality, the United States has advanced a comprehensive executive order called "Ensuring the Safety, Reliability, and Trustworthiness of Artificial Intelligence". The order aims to strike a balance between unrestricted development and strict regulation.

Countries worldwide are actively pursuing AI strategies to cope with the LLM revolution. Russian President Vladimir Putin recently announced a new Russian AI development strategy to counter Western influence in the technology. He is a bit late to the party, as the US, China, the UK, and other countries have already made significant progress on this path. It is also peculiar that he is launching this plan now, considering he famously said in 2017 that the country leading in AI "will become the ruler of the world".

OpenAI and its Mysterious Q*

Now, there is a new mystery surrounding OpenAI. The enigmatic Q project (pronounced "Q-star") has become the next big news. Researchers have named this project "Q" after the quartermaster, the top brain behind creating gadgets for James Bond in the movies.

According to Reuters, the OpenAI board received a letter from researchers about the progress of this hitherto unknown project just days before they fired Altman. The letter warned the board that Q* could pose a threat to humanity.

There is speculation that the board was unaware of Q*, which could be a primary reason for Altman's dismissal. However, this possibility seems unlikely as Ilya Sutskever is both the Chief Scientist of AI and a board member. Supporting this, the platform reports, "I can report that the board has never received any such letter about Q*".

The Path to AGI?

Now, there are various rumors about what Q* could be: a new neural-symbolic architecture (which would be a significant development) or a more moderate but still impressive combination of existing LLMs with several known technologies, resulting in something better than the current state of the art.

An effective neural-symbolic architecture does not yet exist on a large scale, but such systems could enable AI to learn with less data while better explaining their behavior and logic. Several companies and academic institutions, including IBM, are developing such systems, with IBM considering this architecture as a "pathway" to achieving Artificial General Intelligence (AGI). AGI is still not clearly defined but is generally seen as the ability to perform tasks at or beyond human levels of capability at machine speed and information processing.

As reported by The Atlantic, Q is likely not yet at this neural-symbolic breakthrough. Nonetheless, if Q's progress can enter the market, it will be another step towards AGI. By the way, NVIDIA CEO Jensen Huang has stated that AGI could be achieved within five years.

Microsoft President Brad Smith has a slightly different perspective, as reported by Reuters: "You will not see so-called AGI, where the computer is more powerful than humans, in the next 12 months. That will take years, if not decades...".

As we have seen, breakthroughs like ChatGPT and the technologies that potentially power systems like Q* have sparked a wave of optimism, concern, regulation, competition, and speculation. The rapid progress in artificial intelligence over the past year is not just a technological milestone but also a reflection of our relentless pursuit of knowledge and mastery over our own creations.

The coming year will undoubtedly be as exciting and nerve-wracking as the past year. Where we are headed depends on how successfully we harness this energy and guide it.