The Daily AI: Issue #16

"AI agents are the future business partners: tireless, insightful, and without the need for coffee breaks."

Welcome to The Daily AI Show Newsletter, your deeper dive into AI that goes beyond the latest news. In this issue:

• Discover how to use multimodal LLMs to supercharge your creative problem-solving

• The fastest way for a business to leverage the new power of 01-preview

• Learn how the pros are using AI music tools and how you can capitalize on their discoveries

Plus, we discuss California’s new AI laws, Salesforce’s new Agentforce and how to get free training, better tools to help you build your YouTube presence, and a lot more.

It’s Sunday morning.
Sleeping-in is overrated.
There is no better time to level up.

We got you!
The DAS Crew

How Multimodal AI Enhances Creative Problem-Solving

Multi-modal AI, which integrates various forms of input like text, images, and audio, is transforming problem-solving and creative processes.

By leveraging different data types, these models provide a richer context and generate innovative solutions that go beyond what single-mode AI can achieve.

For example, integrating visual data with text enables users to share screenshots of workflows or error messages and get immediate, relevant responses. This capability can be invaluable for troubleshooting and creative ideation, saving time and reducing frustration.

One of the exciting aspects of multi-modal AI is its potential to enhance collaboration and creativity.

Imagine discussing a project out loud while sketching ideas on a digital whiteboard. The AI could analyze the spoken words and visual cues in real-time, providing feedback and suggestions to refine the concept further. This dynamic interaction opens up new possibilities for creatives, engineers, and professionals in various fields.

Multi-modal AI also simplifies complex tasks like coding or project management. By offering contextual support across different modes of communication, it can identify patterns and provide insights that would be hard to spot otherwise. This makes it a powerful tool for both individual and team-based problem-solving.

#WHY IT MATTERS

• Enhanced Problem-Solving: By combining text, images, and audio, multi-modal AI can address complex issues more comprehensively, making it easier to find effective solutions.

• Improved Collaboration: Multi-modal interactions allow teams to work together more seamlessly, using a combination of spoken words, visual aids, and text inputs.

• Creativity Boost: AI-driven feedback during brainstorming sessions can help users overcome creative blocks and develop more refined ideas.

• Simplified Workflow: Multi-modal capabilities streamline processes by offering more intuitive interactions, reducing the need for repetitive tasks and manual data entry.

• Broader Accessibility: These tools democratize access to advanced AI, enabling people from various backgrounds and skill levels to leverage its potential effectively.

Beyond the Hype: How Businesses Can Leverage OpenAI's 01-Preview

The 01-Preview model from OpenAI is garnering attention for its ability to tackle complex reasoning tasks and provide strategic insights.

However, businesses looking to integrate this model need to be strategic about how they use it. While it shows impressive capabilities in coding and mathematics, its true strength lies in the ability to perform deeper, more nuanced analysis that previous models couldn't manage as efficiently.

For companies in sectors like sales, strategic planning, and operations, 01-Preview can go beyond surface-level insights to offer detailed plans and recommendations.

One notable use case is its potential to draft complex outbound sales strategies that balance efficiency with personalization. This is especially critical in industries facing communication saturation, where prospects are increasingly using AI filters to manage outreach. By employing 01-Preview, sales teams can refine their approaches to focus on AI-resistant engagement techniques and strengthen their personalization efforts.

That said, there are limitations to be aware of.

01-Preview currently lacks the ability to upload documents or access the internet, which can restrict its usefulness for certain types of research or real-time data retrieval tasks. Additionally, with a cap on the number of interactions allowed, businesses need to be judicious in how they deploy this tool, often starting with GPT-4 to fine-tune prompts before escalating to 01-Preview for deeper analysis.

#WHY IT MATTERS

• Advanced Reasoning Capabilities: Unlike other models, 01-Preview can offer comprehensive, multi-step analysis, making it ideal for businesses that require deep strategic insights.

• Sales Strategy Enhancement: It provides the ability to draft complex, adaptive sales strategies that can stand out in an AI-saturated communication landscape.

• Limited Usage: The model's current restrictions, such as a lack of document upload and limited interactions, mean that businesses must plan carefully to maximize its benefits.

• Strategic Implementation Required: For companies to get the most out of 01-Preview, it should be integrated thoughtfully, with initial prompt refining done through GPT-4 to conserve interactions.

• Potential for Complex Use Cases: Industries like drone sales and large-scale operations can benefit significantly from 01-Preview's advanced reasoning abilities to solve intricate business challenges, such as optimizing product recommendations based on nuanced variables.

How Music Pros Are Leveraging AI to Finish Masterpieces

With tools like Suno and Udio, AI-generated music is no longer just a novelty—it’s becoming a powerful collaborator for both amateur and professional musicians.

Artists like Timbaland are exploring how AI can take a basic melody or beat and transform it into a full track. As he mentioned, AI can now interpret the producer’s cues to add elements like vocals and additional layers of instrumentation, all without the need for a live band or extensive studio time.

The implications for the music industry are significant.

AI allows musicians to experiment more freely, trying out different genres and arrangements without committing to costly recording sessions. For artists who may not have the resources of a Timbaland, this could democratize music production, allowing anyone to create polished tracks with just a basic understanding of how to guide the AI.

Beth noted the improvisational nature of working with AI, comparing it to a “musical partner” that can expand on your initial ideas in unexpected ways. This collaborative potential opens up new creative possibilities, making AI a valuable tool for artists looking to push the boundaries of their sound.

#WHY IT MATTERS

• Creative Collaboration: AI can act as a co-creator, adding elements like vocals, instruments, or even new melodies to a track, helping artists experiment with new sounds.

• Cost Efficiency: AI reduces the need for expensive studio time and live musicians, making high-quality production more accessible.

• Democratization of Music: Independent artists can now produce professional-grade music without a big budget, leveling the playing field in the music industry.

• Rapid Iteration: AI enables quick iterations of songs, allowing artists to experiment with different versions and arrangements before settling on the final track.

• Future of Music Production: As AI tools continue to improve, they will likely become a staple in the toolkit of not just mainstream artists, but also indie musicians and content creators looking for efficient ways to produce original music.

Just Jokes

Spotter Studio Gets Called Up From The Bench

Brian, Jyunmi, and Beth got together to start testing out Spotter Studio on a 60 day free trial.

What is it?

A website that helps YouTube creators (like us) brainstorm, research, and plan videos that connect with their audience.

We will report back as we use it more and hopefully you will see even better shows in the near future.

Free AI Training and Unmanageable Agents

Karl and Beth were talking about the recent Salesforce Agentforce announcement. Karl was wondering if future employees on top of either their SFDC role or part of marketing ops roles. He wondered about a future where every SaaS has agents and eventually the entire system becomes too complicated to manage.

Beth said the impression she got from the recent SFDC announcement about offering free AI training in Trailhead was about helping reskill the SF Pro workforce. As an example, she said some instructor-led courses were also free that typically cost $900.

If you want to learn more about the free AI trainings, head over to this blog.

 

A World Where AI Is Always Working On Hard Problems

We talk a lot in our Slack channels about what we think the future looks like for businesses. One reoccurring theme is the idea that in the future we will all have agents that can think about hard problems for much longer than a few seconds. We imagine a near future where all businesses can ask AI to go away and think hard about how to solve a problem. Then it will return at some point in the future with its proposed solution.

As Noam suggests in his tweet below, the cost might be high to have an AI that is “on” for long periods of time.

But what if the solution saves the business?
Saves jobs?
Lowers OJI’s?

Would that cost be too high?

Did you know?

“The Pile” is a 112 GB dataset that many LLMs have been trained on.

It was created by EleutherAI in 2020 and is composed of 22 smaller datasets. Instead of a large internet crawl, this dataset is unique because it is a curated selection of data chosen by researchers at EleutherAI to contain information they thought language models should learn.

While massive, it is only 4% of text data available on the internet.

Fancy building your own LLM? You can download The Pile here.

This Week’s Conundrum
A difficult problem or question that doesn't have a clear or easy solution.

The AI Art Repatriation Conundrum:

AI can now create perfect replicas of famous artworks, making it possible for digital or physical reproductions to be displayed in galleries worldwide. This could expose millions to masterpieces that are currently held in distant museums, often far from their cultures of origin due to historical theft or colonialism. These AI-generated pieces could democratize art, bringing cultural treasures back to their communities and allowing billions to experience their beauty. However, this also risks diminishing the uniqueness and value of the original artworks and the prestige of the museums that house them.

The conundrum:
Should AI-generated replicas of culturally significant artworks be displayed in galleries around the world, reconnecting them with their cultures and democratizing access, even if it diminishes the value and prestige of the original pieces? Or should we protect the exclusivity and cultural capital of the originals, leaving the art in its current locations, despite the injustices and limitations on accessibility?

Google Introduces Data Gemma to Combat AI Hallucinations

Google’s latest innovation, Data Gemma, aims to tackle the persistent issue of AI hallucinations by using the Data Commons as a ground-truth reference. This approach leverages real-world, curated data to cross-check and validate AI-generated content. Andy, reporting from Tokyo, noted, “It’s a significant step forward in improving the reliability of AI models. Data Gemma could fundamentally change how we trust AI-generated responses”.

Neuromorphic Computing Breakthrough from University of Limerick

The University of Limerick, in collaboration with an Indian university, has made strides in neuromorphic computing, achieving an efficiency of 4.1 tera operations per second per watt. This development brings us closer to creating eco-friendly, brain-inspired computing systems that could be embedded in everyday materials. Jyunmi couldn’t hide his excitement, comparing the potential applications to something out of Minority Report. “Imagine AI integrated into your cereal box or clothing—it’s almost sci-fi becoming reality!”

Snap's New AR Glasses: Practical or Gimmick?

Snap unveiled their new Spectacles, complete with a sleek design and AR capabilities, but reactions are mixed. “They look like snorkel goggles!” Jyunmi quipped, highlighting the aesthetic challenge. While Snap's push into AR is commendable, the practicality of wearing such oversized glasses daily is still up for debate. However, if priced affordably and made more discreet, these glasses could find a niche audience among tech enthusiasts​.

Salesforce Launches Agentforce, Aiming to Revolutionize CRM

Salesforce has introduced Agentforce, an AI-powered agent that can analyze data and execute tasks like scheduling complex installations. Unlike previous solutions, this AI is deeply integrated into Salesforce’s ecosystem, making it highly customizable for enterprise needs. “This is big,” Beth explained, “because it’s not just automation—it’s a move towards a more intelligent, self-sufficient CRM. But expect the name to change to something like 'Einstein Agent' by next Dreamforce”​.

California’s New AI Laws Target Deepfakes and Actor Rights

California has passed a series of AI-related laws, including one requiring platforms to label or remove AI-generated deepfakes related to elections and another mandating studios to obtain permission before creating digital replicas of actors. The state also prohibits using digital likenesses of deceased actors without consent from their estates. Jyunmi pointed out, “With these laws, California is setting the precedent for AI governance that could influence national and even global standards”​.

Snap Adds Generative AI and Lens Creation Features

Snap has expanded its AI offerings with new tools, including AI-powered Lens creation and generative text-to-video capabilities. Users can now design custom lenses and create 3D models with just a text prompt, pushing the boundaries of AR creativity. Jyunmi commented, “This is a smart move for Snap. They’re making it easier for users to engage with AR, and it positions them as a leader in consumer-facing AI tech”.

Microsoft Copilot Enhancements: More Than Just a Rebrand?

Microsoft’s recent updates to Copilot include features like Copilot Pages, PowerPoint's narrative builder, and Excel enhancements that hint at capabilities similar to a code interpreter. “It sounds promising, but the real test will be in how seamlessly these tools integrate into everyday workflows,” Karl noted. While Copilot’s capabilities are expanding, its adoption will depend on how well users can leverage these new tools without a steep learning curve​.

Did You Miss A Show Last Week?

Enjoy the replays here on YouTube or take us with you in podcast form on Apple Podcasts or Spotify.