The Daily AI Show Newsletter Issue #2

Welcome to Issue #2 of our newsletter.

We’re glad you’re here. If you don’t know, The Daily AI Show is a live weekday show at 10am Eastern on YouTube and LinkedIn. Friday was our 225th show and we just passed 1k+ subscribers on YouTube.

This newsletter is our way of sharing more AI insights that don’t always make it on the live show.

In this edition:

• Does the world really know about ChatGPT?

• AI now stands for Apple Intelligence

• Frontier Models Have No Use For Moore’s Law

We also dig into AGI a bit, talk about the never-ending AI waiting game for new releases and discuss an AI tool that is blowing our crew away this week.

Let’s go!

-Andy, Beth, Brian, Eran, Jyunmi, Karl & Robert.

Is Reuters Right About Generative AI?

On Monday’s show, we discussed a recent report by Reuters titled "AI and the Future of News." The report surveyed approximately 2,000 people in each of six different countries to understand their awareness and usage of generative AI in news and journalism. We explored the public's perception, the level of recognition of various AI tools, and the potential biases introduced by the survey's scope.

The main takeaway is that while tools like ChatGPT are gaining some recognition, a significant portion of the surveyed population is still unfamiliar with many generative AI tools. The survey highlights a gap between AI awareness in tech circles and the general public’s understanding.

WHY IT MATTERS

• Public Awareness Gaps: Despite the rise of AI tools, a substantial portion of the population remains unaware of their existence and capabilities. This indicates a need for more education and outreach to increase public awareness of AI capabilities and bridge the gap between technologists' knowledge and what the public understands.

• Differing Usage Patterns: The report shows that AI is used more frequently in personal contexts than in professional ones. This suggests that while people are experimenting with AI for entertainment or minor tasks, there is still a significant opportunity for businesses to integrate AI more effectively into their operations.

• Regional Differences: The survey highlights regional variations in AI tool recognition, with countries like the US and UK showing higher awareness compared to other nations (except Denmark!). This suggests that AI education and adoption strategies might need to be tailored to regional contexts to be more effective.

• Education and Training: The findings emphasize the need for improved AI literacy. As AI tools become more prevalent, both individuals and organizations must be better equipped to understand and utilize these technologies effectively.

What Is Apple Intelligence?

Apple is focusing on integrating AI capabilities directly into their devices with minimal reliance on external platforms like ChatGPT.

Notably, Apple Intelligence will be available in beta on the latest iPhone models and devices with Apple’s M1 chip or later, as part of iOS 18, iPadOS 18, and macOS Sequoia.

WHY IT MATTERS

• Device-Centric AI: Apple’s approach of embedding AI directly into devices emphasizes privacy and security, as AI operations occur locally rather than relying heavily on cloud-based services. This could enhance user trust by ensuring data remains on the device

• Incremental Innovation: Apple’s strategy focuses on making AI a feature rather than a standalone product. This allows for gradual integration into users' daily lives, aligning with Apple’s philosophy of delivering practical enhancements rather than revolutionary overhauls.

• Market Differentiation: By not directly competing with frontier AI models like ChatGPT or Gemini, Apple differentiates itself through a unique value proposition that highlights device integration and user experience rather than sheer AI capabilities.

• Future AI Integration: Apple's modular approach, allowing Siri to interface with external AI systems like ChatGPT, provides flexibility. This means users can benefit from both Apple’s localized AI processing and the broader capabilities of external AI platforms when necessary.

Can Frontier AI Models Keep Growing at 5x per Year?

Thursday we talked about EpochAI.org’s recent study called “Training Compute of Frontier AI Models Grows by 4 to 5X Per Year”. It was a discussion about the rapid advancements in training compute for AI, how it's measured, and what the future might hold for the expansion of AI capabilities.

WHY IT MATTERS

• Exponential Growth: The training compute for frontier AI models has grown 4-5x annually, far surpassing Moore’s Law. This trend has driven a dramatic increase in AI capabilities, and if it continues on this pace it will lead to ever more powerful and sophisticated AI applications in various fields.

• Sustainability Concerns: The rapid increase in compute demands raises significant sustainability issues. The environmental impact of training large AI models is substantial, emphasizing the need for more energy-efficient practices and technologies in AI development.

• Future-Proofing AI Strategies: As AI models continue to grow in complexity, businesses must plan for the evolving nature of AI capabilities. Investing in adaptable AI solutions today can ensure that businesses remain competitive as AI technologies advance.

• Holistic Ecosystem View: Understanding the entire AI ecosystem, including hardware limitations, sustainability, and efficiency improvements, is crucial for businesses. This broader perspective helps in making informed decisions about AI investments and deployments.

• Disruptive Technology Developments: Advances in training methods, algorithms, model design and processing hardware technology will allow training compute costs to come down, without impairing the further 4-5X advancement of AI technology capabilities.

JUST JOKES

Keeping Us Up at Night:
How Quickly The Pace Towards AGI Might Change

“I can see how AGI will be built. It’s no longer about estimates of human brain size and hypotheticals and theoretical extrapolations and all that—I can basically tell you the cluster AGI will be trained on and when it will be built, the rough combination of algorithms we’ll use, the unsolved problems and the path to solving them, the list of people that will matter.

Right now, there’s perhaps a few hundred people in the world who realize what’s about to hit us, who understand just how crazy things are about to get, who have situational awareness.”
Leopold Aschenbrenner, a former OpenAI safety and superalignment researcher.

_{Credit: This LinkedIn Post}

Did You Know?

Generative AI can dream up entirely new molecules that have never existed in nature, potentially revolutionizing drug discovery and material science.

Fact Details:

• Generative AI, specifically models like Generative Adversarial Networks (GANs) and Variational Autoencoders (VAEs), are being used to create new molecular structures.

• AI can predict the properties of novel molecules and propose new molecular structures with desirable characteristics for use in medicine (think potential customized drug compounds) or in materials science (like new polymers with custom product attributes).

Example Case:

• Researchers at Stanford Medicine and McMaster University are currently working on 6 novel drugs to combat resistant strains of Acinetobacter baumannii, a leading pathogen in antibacterial resistance-related deaths.

Did You Miss A Show?
Watch The Live Show Replays On YouTube