Text-to-image, text-to-video technologies rock the world
AI models are now capable of generating impressive images (pseudo photos) and videos from text inputs. It’s a big deal. But before we get into that, I’m pleased to say that Museum of AI is targeting a late 2022 opening of its first bricks-and-mortar location, in beautiful downtown Denver! Museum of AI is running a membership campaign ~ plus seeking more advisors, speakers, and civic supporters. We’ve formed a Founder Circle. Drop me a note if you are interested in joining.
Text-to-image takes the world by storm
Today’s opening image was created by Craiyon, with a prompt of “people modern museum entrance”. This next image was generated by DALL-E 2. The prompt was “robot cooking” in celebration of our upcoming show, Cookin’ with AI, Season 2. Pretty interesting. The museum team has been playing with DALL-E for awhile, and its growth is explosive: Around 1,000,000 people were added from the waiting list just this week.
But it’s not all sunshine and puppy dogs
Getty Images, among other marketplaces, is concerned about AI-generated art, especially copyright, legality, and metadata ownership, so has banned these images from their site. It’s still the wild west.
…and video, too? Meta (you might know them as Facebook) announced text-to-video capability. We knew it was coming, but still it’s another impressive milestone in a fast-moving world. (Here’s the research article.) Protocol offers an analysis of troubling side effects: Hiring a deepfake instead of a real person. Gulp. (This might be a good time to tell you about the new Museum of AI Deepfake Detection Kit(TM) ~ please scroll down.)
Experience management gets the AI treatment
Probably no one loves those exit surveys “How was your experience today?”. Conversational AI/chatbots are making headway into experience management and customer support. Laura Casey reported in the Wall Street Journal on some big wins. Take Miami Intl Airport, for example: They can analyze social media feeds to learn more about customers without having to put a survey in front of them. And now they encourage passengers to use their chatbot, which has reduced social media inquiries by half, and customer resolution time “to 10 minutes, from as high as 90.”
AI-automated writing also on the rise
LLMs (large language models) have gained a pretty solid grasp on the English language. Using the OpenAI API playground, we asked GPT-3 for a Museum of AI mission statement. Not bad! “The Museum of Artificial Intelligence is dedicated to providing an immersive and educational experience for all visitors. We strive to create an environment where people can learn about the history and future of artificial intelligence, and explore the ethical and social implications of this technology. We also offer an escape room experience for those who want to test their knowledge.”
For the record, our actual mission statement is “Demystify AI for everyone.”
Finally, a sneak peek at the new Museum of AI Deepfake Detection Kit(TM). A handy little gizmo we’re offering as a perk during our fall membership drive.
And just for fun: Voyager technology recently improvised jazz along with a live band. But the cool part is, it dates back to 1987.