OpenAI’s new o1 models push AI to PhD-level intelligence

[LevelUP]

ChatGPT had severe limitations in things such as math, coding, and PHD level responses. Now it looks like a new model addresses some of these concerns. 

OpenAI’s new o1 models push AI to PhD-level intelligence
https://www.fastcompany.com/91189817/ope...telligence

[bjcheung77]

Is this the ChatGPT 4o Mini? I haven't used it yet... The only thing I actually did with that was ask a couple of questions...

[LevelUP]

Is this the ChatGPT 4o Mini? I haven't used it yet... The only thing I actually did with that was ask a couple of questions...

It's both the o1-preview and o1-mini.

Introducing OpenAI o1-preview
https://openai.com/index/introducing-openai-o1-preview/

Learning to Reason with LLMs
https://openai.com/index/learning-to-reason-with-llms/

[bjcheung77]

My query on ChatGPT 4o mini... List all guaranteed admission program medical schools in USA

In the U.S., several medical schools offer guaranteed admission programs, typically through partnerships with specific undergraduate institutions. Here are some notable ones:

1. Boston University School of Medicine - Offers a guaranteed admission program for students from certain partner colleges.
   
2. Drexel University College of Medicine - Has a combined degree program that provides guaranteed admission for students meeting specific criteria.
   
3. The George Washington University School of Medicine and Health Sciences - Offers a guaranteed admission program for select undergraduates.
   
4. Northeast Ohio Medical University (NEOMED) - Has a BS/MD program with guaranteed admission for students from partner colleges.
   
5. Tulane University School of Medicine - Offers a combined degree program with guaranteed admission for qualified students.
   
6. University of California, Riverside, School of Medicine - Offers a program for students from certain California undergraduate institutions with guaranteed admission.
   

Always verify the details and requirements on the official medical school websites, as programs can change and may have specific eligibility criteria.

Here are a few medical schools that offer guaranteed admission programs:

1. Cleveland Clinic Lerner College of Medicine - Offers programs for select undergraduate institutions, including guaranteed admission pathways.
   
2. Northeast Ohio Medical University (NEOMED) - Has several combined BS/MD programs that include guaranteed admission for students from specific undergraduate institutions.
   
3. The University of Toledo College of Medicine and Life Sciences - Offers a guaranteed admission pathway for select students from partner colleges.
   

For accurate and detailed information, it's always best to check the specific medical schools' websites or contact their admissions offices directly.

And then I followed up with this one... List of easiest medical schools to get into in the USA

While the competitiveness of medical school admissions can vary from year to year, some medical schools are generally considered to have higher acceptance rates or less stringent requirements. Here are a few medical schools that are often viewed as having relatively easier admission processes:

1. University of Mississippi Medical Center
   
2. Louisiana State University Health Sciences Center
   
3. University of Arkansas for Medical Sciences
   
4. East Carolina University Brody School of Medicine
   
5. University of New Mexico School of Medicine
   
6. Meharry Medical College
   
7. Morehouse School of Medicine
   
8. University of Texas Rio Grande Valley School of Medicine
   

Keep in mind that "easier" does not mean easy; these schools still require a solid academic background and commitment to medicine. Always check each school’s specific admission criteria and statistics, as they can change annually.

[LevelUP]

It might sound simple, but one thing ChatGPT had trouble with was reviewing its responses for accuracy before generating a bunch of text.

For example, if you asked ChatGPT a simple question like, "How many characters will be in your response?" it couldn't give you an accurate answer until the o1-preview update.

I encountered similar issues when I wanted a response restricted to a specific set of conditions. It wouldn't give me what I asked for, and I’d have to provide another prompt to fix it. It would say something like, "I'm sorry, here's the fix," but the fix still wasn’t accurate. However, this new version finally works as expected.

Given that ChatGPT can take up to around 2 minutes to process certain queries, I wonder if this level of precision and depth will one day become a premium feature for researchers, possibly costing $1,000 or more?

[vetvso]

It also has issues doing some technical analysis. Website optimization SEO also has issues. I have been able to get some simple scripts created that worked fine. I have not written a whole software program with it yet. But I’m curious if creating specific apps for my website is possible.

[ifomonay]

Right now ChatGPT will invent non-existent sources when you ask it to cite them. "PhD" level means proper credit to the source of its information. I'll believe it when I see it.

[LevelUP]

It also has issues doing some technical analysis. Website optimization SEO also has issues. I have been able to get some simple scripts created that worked fine. I have not written a whole software program with it yet. But I’m curious if creating specific apps for my website is possible.

o1-preview and o1-mini can do all that.  Are you using those or the free version?

Right now ChatGPT will invent non-existent sources when you ask it to cite them. "PhD" level means proper credit to the source of its information. I'll believe it when I see it.

The O1 model represents a huge leap in intelligence that some thought would take 4–10 years to achieve. Here is a graph showing how substantial the leap is.

[NotJoeBiden]

How can an IQ test be valid if the AI model was likely trained on IQ tests questions?

[LevelUP]

How can an IQ test be valid if the AI model was likely trained on IQ tests questions?

You bring up a good point.

In the article, they did talk a bit about this.
"He noted that the Mensa IQ test used for this assessment was designed specifically for offline use and is not included in any AI training data."

If you really want to dig deep into the data, you can read this article from the person who ran the test on AI, which was linked in the other article above.
https://www.maximumtruth.org/p/massive-b...telligence

Also, you can view the questions from the Mensa Norway test
https://www.maximumtruth.org/p/massive-b...telligence

I'm not saying the test was perfect, but it's just one way to validate that there were some significant improvements in intelligence from the previous models.