OpenAI marked its second year of ChatGPT with significant achievements and challenges, setting the stage for an even more ambitious 2024. The company, which launched its first version of ChatGPT on November 30, 2022, quickly became a symbol of the burgeoning field of generative artificial intelligence (genAI). As 2024 unfolds, OpenAI has introduced a range of new developments, from advanced AI models to an innovative search engine, all while grappling with internal tensions and legal battles.
In 2024, OpenAI released notable advancements, including GPT-4, which significantly enhanced performance across text, voice, and vision tasks. Additionally, the company launched o1, a new series of models designed to tackle complex challenges in fields like science, coding, and mathematics. A few weeks ago, OpenAI also introduced SearchGPT, a browser extension that offers “fast, timely answers” directly from web sources, bypassing traditional search engines altogether.
Despite these successes, OpenAI has faced internal struggles. Co-founder Ilya Sutskever resigned, and the company’s team focused on researching superintelligence was disbanded. Furthermore, OpenAI is dealing with multiple lawsuits from U.S.-based news organizations over alleged copyright infringement.
Looking ahead, OpenAI has its sights set on a “giant breakthrough” for 2025. During a Reddit AMA in October, founder Sam Altman and his team outlined their plans to develop “AI agents”—autonomous systems that can perform tasks independently. These agents would allow companies to automate complex processes using large language models (LLMs), which OpenAI’s competitors, such as Google Cloud’s Vertex AI agents, are already working on. However, opinions on these agents remain divided, with some seeing them as a game-changer, while others express concerns about giving AI too much control.
As for upcoming releases, Altman hinted at exciting developments, although he clarified that there would be no “ChatGPT-5” this year. The priority is on refining existing models like GPT-4 and o1. Reports suggest that OpenAI’s next model, Orion, may arrive in December for select partners. This model is expected to be more powerful than its predecessors, with CEO Tadao Nagasaki claiming it will be “100 times more powerful” than GPT-4.
In addition to AI agents, OpenAI is working on Sora, an AI text-to-video model, although its release has been delayed. The company is also refining the next version of DALL-E, the image-generating software, but a release date remains uncertain.
Looking to the future, experts like Kate Devlin from King’s College London warn of potential industry limitations, such as compute and energy shortages. OpenAI may consider scaling down to smaller, more resource-efficient models, possibly focusing on specialized applications in fields like law or healthcare.
As OpenAI prepares for its third year, the company faces both immense opportunities and challenges, from the development of autonomous AI agents to navigating a rapidly evolving industry. The next steps for ChatGPT and its parent company will shape the future of artificial intelligence.
