OpenAI CEO Sam Altman says GPT-5 will be miles ahead of "slightly embarassing at best" GPT-4

Sam Altman's Q&A at Stanford University reveals his plans for the future models

During a recent Q&A session at Stanford University, OpenAI CEO Sam Altman provided significant insights into the future of artificial intelligence technologies spearheaded by his company. Describing the current GPT-4 model as “mildly embarrassing at best,”^[1] Altman promised substantial improvements with the upcoming GPT-5.

He emphasized that each iteration of their models aims to surpass its predecessors in intelligence and efficiency, signaling a continual evolution in the AI landscape. Altman's comments reflect a strategic vision where artificial intelligence not only enhances user experiences but also sets new standards in technological prowess.

Moreover, Altman's discussion at Stanford revealed an unwavering goal to advance AI technology quickly and efficiently. By characterizing GPT-4 as the lowest benchmark for future models, he set the stage for the transformative capabilities of GPT-5, which promises to deliver better performance and accuracy.

GPT-4 was called “the dumbest model”

OpenAI CEO Sam Altman was candid in his criticism of GPT-4 during his presentation, noting it as the “dumbest” model users would ever interact with (he also said it “kind of sucks” back in March).^[2] The shift from GPT-4 to GPT-5 is expected to be significant, with improvements in both performance and accuracy, marking it as a transitional model destined to be overshadowed by its successors. Altman said:^[3]

It's important to ship early and often and we believe in iterative deployment. If we go build AGI in a basement and then the world is kind of blissfully walking blindfolded along, I don't think that makes us very good neighbors

In addition to performance enhancements, the transition from GPT-4 to GPT-5 also aims to address the broader scope of AI’s capabilities. Altman's remarks suggest that GPT-5 will not only outperform its predecessors in terms of raw computational power and task efficiency but also in nuanced areas like natural language understanding and problem-solving skills.

The end goal – AGI

Despite the high operational costs of running ChatGPT, which amounts to $700,000 per day, Altman emphasized that the platform remains ad-free, unlike many competitors such as Meta’s models.

He highlighted that the expenditure on development, whether it be $500 million, $5 billion, or even $50 billion annually, is inconsequential as long as OpenAI stays on track with its ultimate goal of creating Artificial General Intelligence (AGI).^[4]

Altman indicated a high degree of confidence in the forthcoming GPT-5, driven by both technological advances and his profound belief in human ingenuity to leverage these tools effectively, suggesting a transformative impact on how businesses and individuals interact with AI.

Going towards the future

Altman's discussions also touched on the broader societal impacts of AI development. He advocated for AI technologies that align with and propel human progress, emphasizing the need for these systems to evolve in a manner that complements societal advancement.

AI technology has already made significant waves across various industries, sometimes even leading to job reductions due to its efficiency and capabilities.

It has grown to a point where it can realistically mimic human voices, generate visual content from text prompts, and create videos through models like Sora.^[5] Although these advancements might threaten traditional jobs, such as those of voice actors, they also present new opportunities for creative and beneficial applications of AI.

Therefore, the release of GPT-5 is seen as a step toward more complex AI interactions that have the potential to transform industries, reinvent user experiences, and improve public and private sector decision-making processes, rather than just a technological advancement.

In addition to improving AI's functionality, the CEO sees GPT-5 and beyond as a time to incorporate ethical issues into the technology's advancement. This entails encouraging transparency, making sure AI systems are created with an understanding of their possible societal impacts, and creating an atmosphere where technology advances in step with human values.