Mar 15, 2023 13:42 UTC
Mar 15, 2023 at 13:42 UTC
ChatGPT v4 aces the bar, SATs and can identify exploits in ETH contracts
GPT- 4 completed numerous of the tests within the top 10 of the cohort, while the original interpretation of ChatGPT frequently finished up in the bottom 10%.
GPT- 4, the rearmost interpretation of the artificial intelligence chatbot ChatGPT, can pass high academy tests and law academy examinations with scores ranking in the 90th percentile and has new processing capabilities that weren’t possible with the previous interpretation.
The numbers from GPT- 4’s test scores were participated on March 14 by creator OpenAI, revealing it can also convert image, audio and videotape inputs to textbook in addition to handling “ much more nuanced instructions ” further creatively and reliably.
“ It passes a simulated bar test with a score around the top 10% of test takers, ” OpenAI added. “ In discrepancy, GPT-3.5’s score was around the bottom 10%. ”
The numbers show that GPT- 4 achieved a score of 163 in the 88th percentile on the LSAT test — the test council scholars need to pass in the United States to be admitted into law academy.
GPT4’s score would put it in a good position to be admitted into a top 20 law academy and is only a many marks short of the reported scores demanded for acceptance to prestigious seminaries similar as Harvard, Stanford, Princeton or Yale.
The previous interpretation of ChatGPT only scored 149 on the LSAT, putting it in the nethermost 40%.
GPT- 4 also scored 298 out of 400 in the Uniform Bar test — a test accepted by recently graduated law scholars permitting them to exercise as a counsel in anyU.S. governance.
The old interpretation of ChatGPT plodded in this test, finishing in the bottom 10 with a score of 213 out of 400.
As for the SAT substantiation- Grounded Reading & Writing and SAT Math examinations taken byU.S. high academy scholars to measure their council readiness, GPT- 4 scored in the 93rd and 89th percentile, independently.
GPT- 4 bettered in the “ hard ” lores too, posting well above average percentile scores in AP Biology( 85- 100), Chemistry( 71- 88) and drugs 2( 66- 84).
However its AP math score was fairly average, ranking in the 43rd to 59th percentile.
Another area where GPT- 4 was lacking was in English literature examinations, posting scores in the 8th to 44th percentile across two separate tests.
OpenAI said GPT- 4 and GPT-3.5 took these tests from the 2022- 2023 practice examinations, and that “ no specific training ” was taken by the language processing tools
“ We did no specific training for these examinations. A nonage of the problems in the examinations were seen by the model during training, but we believe the results to be representative. ”
The results urged fear in the Twitter community too.
Nick Almond, the author of FactoryDAO, told his 14,300 Twitter followers on March 14 that GPT4 is going to “ scarify people ” and it’ll “ collapse ” the global education system.
Former Coinbase director Conor Grogan said he fitted a live Ethereum smart contract into GPT- 4, and the chatbot incontinently refocused to several “ security vulnerabilities ” and outlined how the code might be exploited
Before smart contract check ups on ChatGPT, its first interpretation was also able at spotting outlaw bugs to a reasonable degree as well.
Rowan Cheung, the author of the AI newsletter The Rundown, participated in a videotape of GPT transcribing a hand- drawn fake website on a piece of paper into code.