[ad_1]
Researchers experimenting with OpenAI’s text-to-image software, DALL-E 2, observed that it appears to covertly be including phrases corresponding to “black” and “feminine” to picture prompts, seemingly in an effort to diversify its output
Expertise
22 July 2022
Synthetic intelligence agency OpenAI appears to be covertly modifying requests to DALL-E 2, its superior text-to-image AI, in an try to make it seem that the mannequin is much less racially and gender biased. Customers have found that key phrases corresponding to “black” or “feminine” are being added to the prompts given to the AI, with out their data.
It’s well-known that AIs can inherit human prejudices via coaching on biased knowledge units, typically gathered by hoovering up knowledge from the web. For …
[ad_2]