- The Intelligence Age
- How AI Can Help Non-Verbal Patients
How AI Can Help Non-Verbal Patients
PLUS: MIT's StableRep uses synthetic images to improve efficiency.
Good morning! Today is Tuesday, November 21st and we're covering OpenAI's path forward under new leadership, an innovative emotional analysis tool for healthcare, and NASA's use of custom GPTs to tap climate data. New to The Intelligence Age? Sign up here.
Drop us a line anytime at [email protected] with feedback.
News & Insights
What’s Next For OpenAI
OpenAI's leadership shakeup has left the AI community spinning. Interim CEO Mira Murati's attempts to stabilize the company amidst a potential mass exodus have had mixed results. As employees threaten to depart for Microsoft, the balance between rapid innovation and safety becomes even more precarious. "OpenAI is nothing without its people," a sentiment shared by many, underscores the importance of human capital in the AI race.
With Emmett Shear stepping in as CEO, OpenAI might shift gears, possibly slowing down to ensure AI safety—a stark contrast to the company's previous tempo. "The way you make it safely through a dangerous jungle at night is not to sprint forward at full speed, nor to refuse to proceed forward," Shear asserts. This careful approach could redefine OpenAI's trajectory and its role in leading generative AI advancements. technologyreview
Why does this matter?
The recent events at OpenAI could signal a broader industry trend towards prioritizing ethical AI development. Could this pivot inspire a new wave of startups focused on transparency and safety, potentially leading to a more responsible AI ecosystem?
A Unique Use Case of Generative AI in Healthcare
In the intersection of AI and healthcare, SeamlessCare emerges as a beacon of innovation. Driven by the personal journey of Dr. Aviva Cohen, this start-up leverages the power of AI to give voice to the non-verbal. From philosophy to digital health pioneer, Cohen's story is one of transformation catalyzed by a profound personal experience—her husband's stroke. Empathy in technology isn't just a concept at SeamlessCare; it's the heart of Empathic, their latest tool that interprets the emotional content of non-verbal vocalizations.
The landscape of AI in healthcare is crowded, yet SeamlessCare distinguishes itself by focusing on the often-overlooked non-verbal cues. Empathic is a testament to the potential of AI to bridge human communication gaps, transcending words to understand emotions. "Empathic enables everyone to understand a non-verbal person even if they have never met before," explains Cohen. With the launch of Empathic and upcoming AI Services, SeamlessCare is not just filling market gaps but also creating new pathways for inclusion in healthcare. siliconrepublic
How NASA’s Putting Custom GPTs to Use
Harnessing NASA's Power API, OpenAI's recent venture into GPTs has delivered a breakthrough in creating AI agents that can access and analyze climate data effortlessly. With a no-code approach, the barrier to entry for crafting sophisticated AI assistants has been significantly lowered. The integration of NASA's Power API with OpenAI's GPTs enables real-time retrieval and interpretation of environmental data. Despite its intuitive design, setting up external API communication poses a challenge, particularly when the API lacks a schema. Moreover, the cost implications of using GPTs remain hazy, especially since usage caps could hinder widespread deployment, and OpenAI might make product changes given their recent leadership turmoil. towardsdatascience
Together with Contentful
The future of intelligent composable content.
As the leading intelligent composable content platform, Contentful enables developers and marketers alike to easily deliver compliant on-brand experiences and speed and scale—all within one unified content system.
With Contentful, you can create infinitely and publish instantly.
Please support our sponsors!
The 5 Pillars Of Trustworthy LLM Testing
In the relentless march of AI progress, trust is the cornerstone. Trustworthy LLMs—not just smart, but reliable and ethically sound. This isn’t rhetoric; it's a necessity. With LLMs like ChatGPT and Bard now ubiquitous, we face the challenge of testing these behemoths. Kolena's insights offer a framework: hallucination, bias, reasoning, generation quality, and model mechanics. Each pillar pivotal, each failure mode a potential pitfall
Yet, diving deeper, it’s the interplay between these pillars that's crucial. Improvement in one area may inadvertently affect another. For instance, enhancing reasoning could expose biases, and focusing on mechanics might impact generation quality. We must balance these pillars to tailor LLMs to specific industries and uses. What's at stake? Potentially, the integrity of outputs in sensitive fields like healthcare and law. towardsdatascience
Around The World
Tech Industry Scrambles to Address Fallout from Turbulent Weekend Impacting OpenAI and Microsoft Operations.
NASA's Power API Fuels the Creation of a Groundbreaking Climate GPT, Leveraging Advanced AI to Predict Environmental Changes With Unprecedented Precision.
OpenAI's ongoing crisis is anticipated to catalyze the emergence of a new wave of innovative artificial intelligence startups, shaping the future landscape of the AI industry.
I was honored to share the TED AI stage with Ilya on Oct. 17. His speech video is out today (mine's still being edited). I think it provides relevant context tokens to the ongoing events. Transcript starting at ~10'20":
As AI continues to progress, as technology advances, [...]… twitter.com/i/web/status/1…
— Jim Fan (@DrJimFan)
Nov 21, 2023
Synthetic Imagery Sets New Bar In AI Training Efficiency
MIT's latest revelation in AI training methods is turning heads. StableRep, a brainchild of the institute's researchers, harnesses synthetic images to outperform traditional training techniques relying on real images. This method relies on a nuanced strategy named multi-positive contrastive learning which, according to MIT's Lijie Fan, "teaches the model to learn more about high-level concepts through context and variance." The result? A training system that delves into the essence of objects beyond mere pixels.
The implications are sizable. StableRep not only simplifies data acquisition but hits a stride toward a new dawn in AI training. "The capacity to produce high-caliber, diverse synthetic images on command could help curtail cumbersome expenses and resources," explains Fan. Yet, not all is rosy. Challenges like slow image generation rates and potential bias amplification lurk, reminding us that this is a step in a journey, not the destination. techrepublic
Written by Isaac R. Ward, Casey Clifton, and Alex Brogan.
Send us feedback at [email protected] and help us provide the best coverage of Artificial Intelligence possible.
Interested in reaching smart business leaders like you? To become an Intelligence Age partner, apply here.