Adobe released a new suite of generative AI tools called Firefly that’s about to be another game-changer in the art world. The suite will offer a text-to-image feature that is being promoted as easier to use than competitors (think more Canva, less Midjorney), and something that’s even more important to most artists — copyrights.
According to Adobe, everything fed to its models is either from the Adobe Stock library, out of copyright, licensed for training, or open domain, meaning that they have the rights to them. This is unlike similar platforms like Lensa AI and DALL-E, which have been accused of scraping the internet for their imagery and thus violating the rights of many artists to their visual identity.
New AI medicine-discovery platform Pharma.AI has developed a cancer treatment program in a mere 30 days, including synthesizing a new drug that it suggests will be more effective than current treatments. It can also predict a patient’s survival rate with more than 80% accuracy. The findings have been submitted for further clinical trials.
Bye bye, tedious test tubes! Instead of the monotonous rounds of chemical trial and error that researchers typically face, AI is slated be the next great tool to rapidly advance the healthcare industry.
A recent report by NewsGuard found that GPT-4 may be able to create more convincing fake news than its predecessor, as evidenced by a study involving human judges.
Though OpenAI touted the newest iteration of the chatbot as 40% more likely to produce factual responses, the test found that it was still better at elevating false narratives with a more authoritative voice, was less likely to challenge debunked claims, and included fewer disclosures.