Is GPT too woke? OpenAI CEO Sam Altman answers Lex Fridman's burning question
Altman went on to explain that there will never be a version of GPT that is completely free of bias, as biases are inherent to the data sets that the model is trained on
- Mar 26, 2023,
- Updated Mar 26, 2023, 11:36 PM IST
In a recent interview with Lex Fridman, OpenAI CEO Sam Altman discussed the controversial question of whether the language model GPT is "too woke" or biased. Altman began by acknowledging the shifting meaning of the term "woke," but ultimately conceded that GPT was indeed biased, and likely always will be.
“Is GPT too woke?” Fridman questioned Altman to which he replied, “Honestly, I barely know what woke means anymore. I dig for a while and I feel like the word is morphed. So I will say I think it (GPT) was too (woke) biased and will always be.”
Altman went on to explain that there will never be a version of GPT that is completely free of bias, as biases are inherent to the data sets that the model is trained on. However, he emphasized that OpenAI has made significant improvements to the model in order to mitigate bias and increase its accuracy.
Also read: ‘Elon is obviously attacking us,’ Sam Altman responds to Elon Musk's criticism
Altman also noted that some of GPT's harshest critics have praised the improvements that OpenAI has made between the GPT-3.5 and GPT-4 versions. He expressed appreciation for critics who display intellectual honesty and acknowledge the progress that OpenAI has made, while also acknowledging that there is still much work to be done.
“We certainly have more work to do but I appreciate critics who display intellectual honesty like that,” Altman said.
Many news outlets with right-wing leanings, around the world, have accused OpenAI of exhibiting liberal bias based on certain example interactions with its ChatGPT.
Also Read
Netflix is going all in on gaming, 40 new games announced for 2023
Parents can be jailed for sharing their child's photos on social media in this country
For Unparalleled coverage of India's Businesses and Economy – Subscribe to Business Today Magazine
In a recent interview with Lex Fridman, OpenAI CEO Sam Altman discussed the controversial question of whether the language model GPT is "too woke" or biased. Altman began by acknowledging the shifting meaning of the term "woke," but ultimately conceded that GPT was indeed biased, and likely always will be.
“Is GPT too woke?” Fridman questioned Altman to which he replied, “Honestly, I barely know what woke means anymore. I dig for a while and I feel like the word is morphed. So I will say I think it (GPT) was too (woke) biased and will always be.”
Altman went on to explain that there will never be a version of GPT that is completely free of bias, as biases are inherent to the data sets that the model is trained on. However, he emphasized that OpenAI has made significant improvements to the model in order to mitigate bias and increase its accuracy.
Also read: ‘Elon is obviously attacking us,’ Sam Altman responds to Elon Musk's criticism
Altman also noted that some of GPT's harshest critics have praised the improvements that OpenAI has made between the GPT-3.5 and GPT-4 versions. He expressed appreciation for critics who display intellectual honesty and acknowledge the progress that OpenAI has made, while also acknowledging that there is still much work to be done.
“We certainly have more work to do but I appreciate critics who display intellectual honesty like that,” Altman said.
Many news outlets with right-wing leanings, around the world, have accused OpenAI of exhibiting liberal bias based on certain example interactions with its ChatGPT.
Also Read
Netflix is going all in on gaming, 40 new games announced for 2023
Parents can be jailed for sharing their child's photos on social media in this country
For Unparalleled coverage of India's Businesses and Economy – Subscribe to Business Today Magazine