AI Crafts: A Leap in AI Reasoning

Generative AI models, particularly LLMs, are constantly challenging the limitations of advanced reasoning.

Author

Serra Ikiz
September 19, 2024

Hi,

It has been almost one year since OpenAI released their newest model ChatGPT4o. It was an impressive update, with web-browsing capabilities and advanced reasoning. A few days ago, they released an even newer version, said to have an IQ of a PhD student. Today we will dedicate the time to catch up with the most intriguing AI reasoning LLMs contender yet, chatGPT-o1.

On September 12, 2024, OpenAI released their o1 models, signaling a major leap forward for AI tackling complex tasks in STEM (Science, Technology, Engineering, and Mathematics). Whether it's coding, debugging, scientific research, or advanced math, the o1-preview and o1-mini models are built to handle it all with greater precision than their predecessors. They are said to shine at step-by-step reasoning, which makes them particularly good at handling detailed, long-form problem-solving.

These models also come with a focus on safety, self-checking facts, and reducing bias, which makes them more dependable in delivering accurate results. If it sounds too good to be true, well here comes the drawbacks. The o1 models come with a higher price tag and run a bit slower than older models. That’s because of their high computational demands. So, if your task involves methodical, multi-step processes, and you need a hand, these models are probably worth it.

© OpenAI

Right now, they’re being rolled out to select users for early access, with a wider release planned soon. However, if you are already a user of chatGPT, you can test these new models out for a couple of tasks. Let me remark though that they do not have any web-browsing capabilities, if you’re wondering.

❝

The current technology that we have is like that very first cellphone with the black and white screen that can only display those numbers…it just didn’t do much…And at the time, [making a phone call] seems great. And then it took us…many decades from that to get to the iPhones we have today…We’re at this barely useful cellphone.

OpenAI CEO Sam Altman

This is part of OpenAI's effort to advance AI for specialized tasks, enabling it to play a larger role in fields like scientific research, software development, and technical problem-solving that demand high precision. As these o1 models become more accessible, they could transform how AI supports professionals and academics, offering deeper, more reliable solutions across industries.

But the pressing question remains, does this mean AI will replace us? If not, will it make us less productive, lazier? Where do we draw the line between human reason and AI reason?

Thank you for joining us in exploring how AI reasoning is transforming the advanced fields of research.

Best regards,
Yasmine