We are excited to bring Transform 2022 back in-person July 19 and virtually July 20 - 28. Join AI and data leaders for insightful talks and exciting networking opportunities. Register today!
Want AI Weekly for free each Thursday in your inbox? Sign up here.
It’s the beginning of Pride month. At VentureBeat, it’s also the beginning of a six-week-long deep dive into AI innovation, culminating in our fourth annual AI Innovation Awards to be presented at Transform 2022 on July 19th in San Francisco.
We are excited about our amazing nominating committee that, along with members of our editorial staff, will help choose our AI Innovation Award nominees in five categories this year.
Innovation, of course, has many definitions and comes in many forms. When it comes to AI, incredible, jaw-dropping innovation is often accompanied by eye-rolling emojis, more questions than answers and even an industry-driven challenge to Elon Musk.
But AI innovation also leads to new efforts to implement best practices to carefully deploy powerful technology, like today’s announcement from Cohere, OpenAI and AI21 Labs that they “have developed a preliminary set of best practices applicable to any organization developing or deploying large language models.”
Let’s dig in.
– Sharon Goldman, senior editor and writer
This week’s AI beat
The concept of innovation, according to Wikipedia, has gone through a variety of definitions over the centuries: The Greek philosopher and historian Xenophon (430–355 BCE) connected it to political action. The term was an “early-modern synonym for ‘rebellion’, ‘revolt’ and ‘heresy.'” Only after World War II did the notion of technological innovation become tied to the idea of economic growth and competitive advantage — popularized by Austrian-born political economist Joseph Schumpeter.
Today, dozens of press releases sail into my inbox every week touting “innovative” AI-driven products and services. But, not surprisingly, even the most innovative efforts don’t often enjoy a straight-up trajectory, but must naturally move through an iterative, collaborative, debate-filled, critique-driven process — which may or may not lead to successful real-world applications. At least not yet.
Take the current competition for the leader of the AI imagery pack, with Imagen, Google’s new text-to-image generator, taking on OpenAI’s DALL-E 2 last week. Each shows off remarkable photo-realism and claims deep-language understanding.
But what to make of this week’s Twitter feather-fluff over some claiming that DALLE-2 has a “secret language?”
The inevitable commentary from “grumpy linguists in NLP” was immediate, followed by a wave of LOL responses claiming that not only does DALL-E-2 likely not enjoy access to a “secret” language but, at the moment, seems to have a tough time with text in its imagery generally:
All joking aside, at VentureBeat we look forward to honoring the best in AI innovation at the AI Innovation Awards at Transform 2022.
Thanks to our nominating committee members:
Tonya Custis, director of AI research, Autodesk
Seth Dobrin, global chief AI officer, IBM
Andrea Huels, head of AI, Lenovo North America
Di Mayze, global head of data and AI, WPP
Stephanie Moyerman, senior director of risk and trust science, eBay
Shawn Wang, chief AI officer, Anthem
And stay tuned for news about the final nominees and winners!