bb
Awesome! Thanks for doing it and I agree about the explanations. I ran a couple official guide questions to see if AI could give some interesting or elaborate answers and sure enough it missed every single one of the three critical reasoning questions I tried 😬🥴
I think it may be better in sentence correction… but I am very surprised and impressed it even knows what data sufficiency is….
I am also very surprised about GRE. Supposedly it can get a pretty high score on the SAT or GRE but what about charts and graphs that you cannot read? Seems a little confusing…
Posted from my mobile deviceNo problem, BB!
It is fascinating that for these past few weeks, the progress goes very fast, and I did not expect at all such AI would be introduced to the public. I do believe that since it has been released despite all the ongoing concerns, academic institutions and AI experts need to work together to spread awareness and knowledge on using AI responsibly and properly.
I believe that soon many more GPTs will be introduced and existing ones will get frequent updates that make these AIs to obtain higher level of reasoning. Who knows maybe in less than 1 year from now, GPT manage to get 750+ on GMAT.
I currently involve in developing something similar but am quite upset with the slow progress. The company that I work for (hint : fruit logo) is very cautious on releasing it despite of huge ongoing public buzz on AI. There are also major hurdles such as convoluted structures and reasoning capability that is not mature enough. (Sorry, I am so excited on everything that is happening on AI and tech recently that I cannot stop writing)

Old AI that is more dependent on pattern recognition and generating output via Standard Prompting is being replaced by the new one that is able to generate output using logical reasoning based on COT Prompting. Therefore, GPT3’s and GPT4’s responses are more human-like compared to those voice assistant’s (Siri, Alexa, Google, etc) responses, which are robotic.