OpenAI's new AI model, o3, has showcased remarkable capabilities in thinking with images. The system can now perform a variety of tasks that were previously considered challenging for AI, such as:

• Identifying the name of the largest ship in an image and predicting its next destination.
• Zooming in on a tiny sign in an image, cleaning up the image, and reading the text on the sign.
• Identifying the movies that were filmed at a specific location, based on an image.
• Finding "Waldo" in an image, and even identifying the chords being played by a guitar hero in a photo.
• Annotating images, such as highlighting issues with small fabric samples.

The AI's memory and learning capabilities are also impressive. It can recall and summarize previous conversations, and then use that knowledge to teach the user something new, such as the behavior of coral larvae and their preference for the natural soundscape of healthy reefs.

Furthermore, the AI has demonstrated its ability to perform research-like tasks. When presented with an incomplete research work, the system was able to analyze the content, zoom in on relevant sections, and provide a plausible completion of the work. This suggests that the AI may be capable of pushing the boundaries of human knowledge and potentially contributing to scientific advancements in various fields.

Its memory is getting better and better. For instance, you can do the following: little AI, read everything that we talked about before, maybe years and years of data, and now — you can ask it to tell you what it knows about you and teach you something that you don't know yet.

In this example, he likes scuba diving and music. o3 knows that. And here comes something mind-blowing. It starts teaching him about coral larvae, underwater baby corals if you will. And they can detect reef sounds. They prefer the natural soundscape of healthy reefs and if they hear them, they are more likely to attach and grow when exposed to them. So, crazy experiment: put loudspeakers underwater, pipe in the sound of a healthy reef, and there you go. They love it. So much so that they swim toward the sound and settle there.

Wow. Imagine someone knowing you so well and teaching you this. If this would be a person, we would say that this person is very kind and thoughtful. And we might say the same thing for a machine. I think that is incredible.

And now that you know about this, you can ask again, it knows that you know this already, and then it can teach you something new. Maybe you are looking for a job as a research scientist, and once again, it knows what you know, but more importantly, it knows what you don't know, and will be able to help you to prepare for an interview.

The new AI system showcased by OpenAI, o3, has demonstrated remarkable capabilities that could push humanity forward. One of the most impressive feats is its performance on the "Humanity's Last Exam" - a challenging test designed by smart scientists from around the world, which no AI could previously answer satisfactorily.

Just a couple of months ago, the best AI could only achieve an 8% score on this test. However, o3 has now managed to reach an astonishing 25% - a threefold improvement in a short period. This is a stunning achievement, and it suggests that o3 may be capable of doing genuine research and pushing the boundaries of human knowledge.

The ability to complete unfinished research work is another remarkable capability of o3. By thinking with images and showing its thought process, the AI system has demonstrated that it can go beyond simply regurgitating information and can actively contribute to advancing scientific understanding.

This breakthrough has the potential to accelerate progress in various fields, such as drug design, crop improvement, clean energy, and longevity research. With o3's ability to push the boundaries of human knowledge, we may soon witness groundbreaking discoveries and innovations that could significantly benefit humanity.

The development of o3 and its impressive performance on challenging tasks is a testament to the rapid progress in artificial intelligence. As we continue to push the boundaries of what AI can achieve, the potential to solve complex problems and drive humanity forward becomes increasingly tangible.

OpenAI's Codex is a remarkable breakthrough that goes beyond just assisting coders. This powerful AI system can help anyone, regardless of their coding experience, to create real-time applications with ease.

One of the standout features of Codex is its ability to understand and respond to natural language prompts. You don't need to write lengthy, complex code to achieve your desired outcome. Instead, you can simply refer to something a fellow user has done on social media, and Codex will generate a real-time app that replicates the same functionality, such as creating beautiful ASCII character art from an image.

This level of accessibility and user-friendliness is truly remarkable. Codex empowers everyone, not just professional developers, to harness the power of coding and bring their ideas to life. Whether you're looking to automate a task, create a custom tool, or explore new creative avenues, Codex is there to assist you every step of the way.

The implications of this technology are far-reaching. By democratizing coding and making it accessible to the masses, Codex has the potential to unlock new levels of innovation and problem-solving. Imagine the possibilities when everyone can easily create their own applications and tools to address their unique needs and challenges.

In the hands of the creative and the curious, Codex has the power to transform the way we approach technology and problem-solving. It's a testament to the incredible progress made in the field of artificial intelligence, and a glimpse into a future where coding is no longer a specialized skill, but a tool that empowers us all.

To improve the credibility of the information provided, I recommend the following approach:

1. For every answer, look for peer-reviewed sources that support the claims made.
2. Rate the credibility of each source on a scale of 0 to 10, with 10 being the highest credibility.
3. Include the source information and credibility rating with the answer to provide transparency and allow for further verification.

This process will help ensure that the information presented is well-supported by reliable, peer-reviewed sources, and will allow the audience to assess the credibility of the sources used.