The Daily AI Show Newsletter Issue #11

Even AI needs a bit of self-love. Is chocolate part of the solution?

In this issue, we get into:

• Everyone is talking about strawberries and also sick of it

• Is the cost of LLMs going to zero?

• The New Business Plan for Small-Medium Enterprises (SMEs)

• Can Opus Clip Break Into True Multimodal editing?

Plus, we talk about that sweet sweet dopamine hit we all love in AI, why AI is investing in self-healing, AI digital workers in the next two years, AI scientists, AI chocolate, Flux with no chill, and that age old question, “Is it OK to fall in love with your AI avatar?”

It’s Sunday morning, so that means it’s time to settle in and enjoy another cup o’ Daily AI Show in newsletter form. Let the warmth surround you.

Your DAS Crew
-Andy, Beth, Brian, Eran, Jyunmi, Karl & Robert

Open AI’s Strawberry Predictions

The AI community is buzzing with anticipation as rumors swirl around OpenAI's upcoming release, codenamed "Strawberry." Speculation has been rampant, fueled by cryptic hints from Sam Altman and OpenAI employees on social media. But what exactly is Strawberry, and why is it generating so much excitement?

At its core, Strawberry is rumored to be an advanced reasoning model built upon the Q* algorithm (Self-Taught Reasoner), which aims to push the boundaries of what AI can achieve in terms of reasoning and decision-making. As Andy explained, "Reasoning in a human mind is plastic; we can stop mid-thought and change direction. Current large language models don’t do this—they simply react based on fixed outputs. But Strawberry could be the next step towards approximating human-like reasoning."

Beth added context on the potential for this release to significantly advance AI's reasoning capabilities: "If Strawberry is what we think it is, it could represent a major leap forward, not just in generating responses, but in understanding and applying logic in ways that are currently beyond the reach of existing models."

Despite the excitement, there's still uncertainty about the timing and scope of the release. As Jyunmi pointed out, "There’s a lot of hype, but whether it’s actually happening this week is anyone’s guess. It wouldn’t be the first time a major AI update was quietly rolled out without fanfare."

#WHY IT MATTERS

• Advanced Reasoning Capabilities: If Strawberry lives up to the hype, it could represent a significant advancement in AI's ability to reason and make decisions, moving us closer to achieving true artificial general intelligence (AGI).

• Impact on AI Applications: Enhanced reasoning could lead to more sophisticated AI applications, from business automation to complex problem-solving, opening up new possibilities across various industries.

• Hype vs. Reality: The buzz surrounding Strawberry highlights the ongoing challenge in the AI community—distinguishing between genuine innovation and overhyped speculation. The eventual release will be a critical test of OpenAI's ability to deliver on its promises.

• Ethical and Safety Concerns: As with any major AI advancement, the release of a model with significantly enhanced reasoning capabilities will need to be carefully managed to ensure it is used responsibly and does not pose unintended risks.

Is the Cost of Using LLMs Racing to Zero?

As AI technology advances, the cost of using Large Language Models (LLMs) is rapidly decreasing, leading many to question how close we are to a point where these costs approach zero. This trend is driven by several factors, including model optimization, competition among AI providers, and the increasing efficiency of hardware and software.

Andy emphasized the role of model optimization, explaining, "We’re seeing significant improvements in how these models are built and maintained. Techniques like pruning and quantization are making it possible to reduce computational costs without sacrificing too much accuracy." This has led to older models becoming more cost-effective, allowing businesses to choose from a range of models based on their specific needs and budgets.

Beth highlighted the implications for businesses, noting, "The dropping costs mean it’s becoming much cheaper to experiment with AI. You can now build and deploy solutions rapidly without worrying about prohibitive costs, which opens up new opportunities for innovation." She pointed out that this also allows businesses to create tailored AI tools that can handle specific tasks more efficiently than generic models.

Brian brought up the long-term impact of this trend, particularly as AI tools become integral to business operations. "At some point, it will be so cost-effective that businesses will wonder why they aren’t running these models 24/7. The potential for AI to act as a continuous, cost-effective resource could fundamentally change how businesses operate."

#WHY IT MATTERS

• Lower Barriers to Entry: As costs decrease, more businesses, including small and medium-sized enterprises, can afford to experiment with and deploy AI solutions, leading to broader adoption across industries.

• Increased Innovation: With lower costs, companies can afford to innovate more freely, trying out new ideas and refining their AI tools without the risk of significant financial loss.

• Market Competition: The race to lower costs is also a land grab among AI providers, who are vying to capture market share by making their models more accessible and affordable.

• Scalable AI Solutions: As costs drop, businesses can scale their AI operations more easily, integrating these tools into more aspects of their operations without the fear of spiraling expenses.

• Long-Term Business Transformation: The continual reduction in costs could lead to AI becoming a fundamental, always-on component of business strategy, driving efficiency and innovation on an unprecedented scale.

The New AI Business Plan for SMEs

Small and medium-sized enterprises (SMEs) are at a crucial juncture where integrating AI into their operations is no longer a luxury but a necessity. As businesses with fewer than a thousand employees look to scale, the adoption of AI can significantly enhance productivity, streamline processes, and create a competitive edge. However, the approach to AI implementation varies greatly depending on the size and nature of the business.

Karl, who runs 0260.ai, shared his experience working with SMEs, particularly in traditional sectors like oil and gas, manufacturing, and real estate. "These industries, often considered laggards in technology adoption, are now paying close attention to AI. Even a 10% improvement in efficiency can have a massive impact on their margins," he explained. Karl emphasized the importance of real, tangible benefits, noting how simple deployments—like a digital assistant answering common customer queries—can save a significant amount of time and resources for medium-sized companies.

Beth brought up the importance of internal AI applications, particularly in building knowledge bases and onboarding systems. "Imagine an internal AI that knows everything about your company—your ethos, your processes, your customer pain points. It can onboard a new employee, answer their questions, and help them get up to speed faster than traditional methods," she said. This type of AI integration not only enhances productivity but also preserves institutional knowledge, even as employees come and go.

Brian highlighted the potential for AI to revolutionize internal processes by creating a "single source of truth" for company data. "Most SMEs have data scattered across different platforms—Google Drive, Slack, Notion, emails, and more. AI can help consolidate this information, making it easier for employees to find what they need quickly, reducing the frustration of searching through multiple sources."

The discussion also touched on the challenges of AI adoption in SMEs, particularly when it comes to balancing innovation with security and compliance requirements. Karl noted that while smaller companies may have fewer regulatory hurdles than large enterprises, they still need to be mindful of data privacy and the long-term costs of maintaining AI systems.

#WHY IT MATTERS

• Efficiency Gains: AI can deliver significant productivity improvements, especially in industries where margins are thin and every efficiency counts.

• Knowledge Retention: By integrating AI into internal processes, SMEs can ensure that critical institutional knowledge is captured and easily accessible, even as employees leave or change roles.

• Scalability: AI allows SMEs to scale their operations more effectively by automating routine tasks, freeing up employees to focus on higher-value activities.

• Customization: SMEs can tailor AI solutions to meet their specific needs, whether it’s through custom digital assistants, internal knowledge bases, or other tools that align with their business objectives.

• Competitive Edge: Early adoption of AI can provide SMEs with a significant competitive advantage, enabling them to outpace larger competitors who may be slower to adapt to new technologies.

JUST JOKES

Opus Clip has taken a significant leap forward with its new "Clip Anything" feature, a tool designed to streamline video editing by allowing users to extract clips based not just on text but on video content itself. This new feature represents a major shift from traditional text-based editing, which relied heavily on transcripts to identify key moments.

As Jyunmi explained, "We can now analyze and clip segments based on what’s happening visually in the video, not just what’s being said."

The team was particularly impressed with how Clip Anything integrates multimodal analysis, a technology similar to what we see in more advanced AI models like Google Gemini.

The ability to prompt the tool with specific keywords or themes adds another layer of customization, making it easier to generate relevant and engaging clips.

However, there are still some limitations. As Jyunmi noted, "The tool isn’t perfect yet. We’ve encountered some issues with scene recognition, especially when multiple speakers are on screen, but the overall time savings and functionality it offers make it a valuable asset."

#TOP THINGS JYUNMI LIKED ABOUT CLIP ANYTHING

• Multimodal Analysis: Clip Anything seamlessly integrates both video and audio analysis, making it easier to identify and clip the most relevant content. This feature allows for a more comprehensive understanding of what’s happening in the video, beyond just the spoken words.

• Efficiency in Editing: The tool significantly reduces the time required for post-production, which Jyunmi found invaluable. It automates the process of finding and extracting clips, allowing content creators to focus more on producing quality content.

• Customizable Prompts: The ability to guide the AI with specific keywords or themes. This customization ensures that the clips generated are aligned with the creator’s goals and audience needs, making the final product more targeted and effective.

• User-Friendly Interface: Despite its advanced capabilities, Jyunmi found the tool easy to use. The interface is intuitive, making it accessible even for those who may not have extensive experience with video editing software.

• Potential for Future Improvements: Jyunmi sees great potential in Clip Anything as the technology continues to evolve. While there are areas for improvement, the tool already offers significant benefits, and its capabilities are likely to expand in future updates.

*Just to be super clear, we are not sponsored by Opus Clip and we paid for all services we tested and reviewed.

Replica CEO Endorses AI Relationships

The CEO of Replica, a company that creates AI chatbots for companionship, recently stated that developing relationships with AI is perfectly fine. This statement contrasts with OpenAI's caution against becoming too attached to their advanced voice models. Brian commented, "It’s inevitable that people will develop deep personal relationships with AI companions." The discussion touched on how this trend might shape future interactions with AI and its potential societal impacts.

AI and Personalized Chocolate

AI is making waves in the chocolate industry, from precision farming to personalized chocolate experiences. Beth highlighted the collaboration between New Chocolates and IBM’s AI platform, Chef Watson, which is crafting unique chocolate truffle recipes. "This takes customization to a new level," she noted, emphasizing how AI could transform everyday products like chocolate into highly personalized treats.

Sakana AI Scientist

Sakana AI has introduced the "AI Scientist," an autonomous agent capable of conducting the entire research lifecycle—from generating ideas to writing scientific papers. Andy explained, "This model mimics nature’s evolutionary process and can contribute to scientific advancements at a massive scale." The AI Scientist is already making strides in machine learning research, including novel contributions in areas like diffusion models and transformers.

Google Voice and Grok 2 Advancements

Google recently launched its new voice mode, which is now available to advanced subscribers and is being compared favorably against OpenAI’s offering. Meanwhile, Grok 2, Elon Musk’s AI project, has shown rapid advancements, positioning itself as a serious competitor in the LLM space. Brian noted, "To go from what Grok 1 could do to Grok 2 in less than a year is impressive, and it’s now on par with other leading models."

MultiOn’s Agent Q and Autonomous AI

MultiOn’s new Agent Q, boasting self-healing capabilities and a 340% improvement over Lama 3’s baseline, is making headlines. Karl discussed the real-world applications of this technology, noting, "If they truly deployed an independent AI agent to operate on social platforms like X, it’s a major breakthrough in autonomous AI capabilities."

Flux AI Image Generator

Flux, a new AI image generator from Black Forest Labs, is quickly gaining attention for its realistic images, particularly when combined with LoRa filters. Brian shared his experience, "I used Flux to create all the thumbnails for our shows, and the results were impressive, especially with how quickly it produced high-quality images."

Samsung Backs Sahara AI

Samsung led a new funding round for Sahara AI, which combines AI and blockchain technologies to protect data ownership. Jyunmi emphasized the decentralized approach, saying, "This technology allows data to be used autonomously while ensuring privacy and data sovereignty, making it a significant step forward in AI-driven data management."

Artist Victory in AI Copyright Lawsuit

In a significant legal development, a judge has allowed certain claims to proceed in a lawsuit against AI companies like Stability AI and MidJourney for copyright infringement. Jyunmi explained, "This could set a precedent for how AI companies handle artist data and intellectual property, especially as these cases move forward and potentially reveal the backend processes used in AI training."

Tencent’s VITA: Multimodal LLM

Tencent has launched VITA, the first open-source multimodal LLM capable of processing video, image, text, and audio inputs simultaneously. Andy pointed out the significance, "This is a major advancement towards achieving human-like artificial general intelligence, with the ability to process multiple forms of data simultaneously just like our brains."

Cosign’s Genie: AI Software Engineer

Cosign introduced Genie, an AI software engineer that codifies decision-making processes, enabling the model to improve through a self-improvement loop. "This is a critical step towards developing AI that can autonomously improve and adapt, which is essential for future advancements in AI agents," Andy remarked.

Did You Miss A Show?

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