Episode #123: Trump AI Policies, Problems with OpenAI’s New Model, GenAI for Business Strategy & Visa’s 500 AI Use Cases

Release Date: November 12, 2024

Hosts: Paul Roetzer and Mike Kaput
Guest Speaker: [Unnamed]

1. Introduction

In Episode #123 of The Artificial Intelligence Show, hosts Paul Roetzer and Mike Kaput delve into a variety of pressing AI topics. This episode navigates through potential shifts in AI policy under President-elect Donald Trump, examines the recent challenges faced by OpenAI’s latest model, explores generative AI’s role in business strategy, and highlights Visa’s extensive implementation of AI across its operations.

2. Trump AI Policies

Timestamp: [00:00 - 06:20]

Overview: The hosts discuss the anticipated changes in AI policy with the incoming Trump administration. They emphasize that while AI was not a focal point during the election campaigns, its significance is undeniable for future governance.

Key Points:

• Dismantling Biden's AI Executive Order: Paul notes, “they didn't know how the public perceived AI… but we knew it was gonna be fundamental to whatever happened… once the administration… came into office” (00:00).

• Influence of Key Advisors: The administration is expected to be influenced by tech leaders like Elon Musk, who has significant financial backing and strong opinions on AI development.

• Regulatory Changes: Potential elimination of the AI Safety Institute established under Biden’s order could lead to less regulatory oversight, fostering a pro-development environment but raising concerns about safety and ethics.

• Trade Policies Impacting AI: Trump’s proposed tariffs on imports, particularly Chinese products, could affect the hardware essential for AI advancements, though the long-term implications remain uncertain.

Notable Quote: Paul reflects on the potential rapid policy changes, stating, “Trump and Elon Musk both have very large egos, very large personalities…and how those two get along for four years is going to be really interesting” (06:20).

3. Problems with OpenAI’s New Model

Timestamp: [17:05 - 29:06]

Overview: The conversation shifts to OpenAI's flagship model, codenamed Orion, highlighting concerns about the slowing pace of advancements in AI models.

Key Points:

• Plateau in AI Development: Reports suggest that Orion shows smaller improvements compared to previous models like GPT-3 to GPT-4. OpenAI employees have indicated inconsistencies in performance despite increased computational costs.

• Scaling Laws Challenges: The traditional assumption that more data and computational power will continuously improve AI models is being questioned. Data scarcity and quality are emerging as bottlenecks.

• OpenAI’s Strategic Shifts: In response, OpenAI is investing in post-training improvements and exploring new reasoning models (O1) to enhance AI capabilities.

Notable Quotes:

• Guest Speaker: “The headline misleading… AI progress is slowing is exactly the opposite of what I'm hearing” (19:13).
• Paul: “These models are getting so much better so quickly… for you, focus on using what we have today to do your plans for next year” (28:23).

Practical Applications: Paul shares a personal use case where ChatGPT significantly accelerated their internal planning process, demonstrating the current utility of AI tools despite discussions about potential slowdowns in model improvements.

4. GenAI for Business Strategy

Timestamp: [29:06 - 39:56]

Overview: Paul and Mike illustrate how generative AI can be leveraged for strategic business planning, using their own experience with ChatGPT to enhance their organization’s growth matrix.

Key Points:

• Accelerating Planning Processes: By utilizing ChatGPT to draft business plans and task lists, the hosts reduced what would have taken hours into minutes, allowing for more efficient meetings and decision-making.

• Enhancing Creativity and Innovation: AI-assisted planning enabled deeper discussions and quicker ideation, leading to actionable strategies that might not have been developed otherwise.

• Integration Challenges: Paul highlights the difficulties in integrating ChatGPT with Google Docs, suggesting potential improvements for seamless workflow integration.

Notable Quote: Mike concurs, “...if we had a really smart employee, like brief us on initial ideas… we would never have gotten there” (34:53).

Takeaway: The hosts advocate for businesses to utilize existing AI tools to unlock immediate value, emphasizing action over apprehension regarding future model advancements.

5. Visa’s 500 AI Use Cases

Timestamp: [39:56 - 44:08]

Overview: Highlighting Visa's extensive adoption of generative AI, the hosts explore how the financial giant has successfully integrated over 500 AI applications to enhance various facets of its operations.

Key Points:

• Diverse AI Implementations: From security tools detecting code vulnerabilities to specialized chatbots acting as subject matter experts, Visa is leveraging AI across multiple departments.

• Strategic Investment: Over $3 billion has been invested in AI and data infrastructure over the past decade to support this expansive AI initiative.

• Governance and Data Protection: Visa has prioritized robust governance frameworks and data protection measures to ensure secure and effective AI deployment.

• Hybrid Workforce Model: The company envisions a future where human employees manage multiple AI-powered digital workers, enhancing productivity and operational efficiency.

Notable Quote: Paul remarks, “This is interesting and it makes sense like a lot on the fraud side that they would have a ton of use cases there” (41:53).

Implications: Visa serves as a benchmark for enterprise-level AI adoption, showcasing how substantial investment and strategic implementation can yield significant operational benefits.

6. Rapid Fire Topics

Timestamp: [44:08 - 68:58]

Overview: The episode concludes with a series of rapid-fire updates covering various AI developments, including OpenAI’s corporate restructuring, domain acquisitions, investment rounds in startups like Perplexity and Physical Intelligence, legal battles, and new AI tools from Anthropic and Google.

Key Highlights:

• OpenAI’s Corporate Restructure: OpenAI is transitioning from a nonprofit to a for-profit entity, engaging with regulators in California and Delaware. This move aims to maintain its social mission while attracting investment, though it faces challenges, especially with Elon Musk’s potential influence.

  Quote: Paul speculates, “if Musk wants his vengeance, this is how you use your newfound influence and power” (45:39).

• Acquisition of chat.com: OpenAI acquired the valuable domain chat.com from Dharmesh Shah, integrating it with ChatGPT. The transaction hints at strategic brand consolidation.

  Quote: Paul shares excitement about the acquisition, saying, “the fact that OpenAI is using it… gives Dharmesh some enjoyment” (47:48).

• Perplexity’s Investment Round: Perplexity is set to secure a $500 million investment led by IVP, valuing the company at $9 billion. Despite growth, the startup faces PR challenges and strategic missteps.

  Quote: Paul notes, “a lot of this stuff where it could use… someone in the room who's been through these things before” (50:21).

• Labor Dispute at The New York Times: Perplexity’s CEO Aravind Srinivas faced backlash for offering AI services to The New York Times during a labor strike, raising concerns about AI replacing human roles.

  Quote: Paul critiques the CEO’s actions, stating, “these self-inflicted wounds are so obvious” (53:37).

• Anthropic’s Quad 3.5 Haiku: Anthropic launched an upgraded AI model that matches larger models’ capabilities while maintaining speed, aiming for widespread adoption with competitive pricing.

  Quote: Paul emphasizes the economic implications, “the models are getting more powerful, they’re getting way cheaper to build” (56:39).

• Amazon’s Investment in Anthropic: Amazon is in talks to invest further in Anthropic, with conditions favoring the use of its own Trainium chips, potentially impacting Anthropic’s flexibility.

  Quote: Paul discusses potential strategic moves, “I can't see Anthropic locking themselves in like 100% to something like this” (58:35).

• Physical Intelligence’s Funding: Robotics software company Physical Intelligence raised $400 million, highlighting the growing investment in foundational robotics software capable of performing everyday tasks.

  Quote: Paul anticipates future advancements, “robotics is going to be a major, major area of investment and progress” (61:15).

• OpenAI’s Legal Victory: A New York federal judge dismissed a lawsuit against OpenAI for using news articles in its AI training, emphasizing that the synthesis of data does not constitute copyright infringement.

  Quote: Paul reflects on legal implications, “What I said was it might have been more explicitly saying that it thinks the labs have run into a plateau on the scaling laws” (63:29).

• Google’s AI Assistant Jarvis Leak: Google accidentally revealed details of its new AI assistant, Jarvis, which can autonomously perform tasks like purchasing groceries and booking flights. The reveal was premature, hinting at future AI capabilities.

  Quote: Paul advises moderation in expectations, “temper our expectations” (65:02).

Conclusion: The rapid-fire segment underscores the dynamic and multifaceted nature of the AI landscape, with significant movements across corporate strategies, legal frameworks, and technological advancements shaping the future of artificial intelligence.

Final Thoughts

In this episode, Paul and Mike provide a comprehensive overview of the current AI environment, touching on policy shifts, technological challenges, strategic business applications, and notable industry developments. Their discussions emphasize the importance of leveraging existing AI tools for immediate business gains while keeping an eye on evolving regulatory and technological landscapes.

Key Takeaway: Businesses should focus on integrating and utilizing present AI capabilities to drive efficiency and innovation, rather than waiting for future model advancements.

For more insights and updates, subscribe to Marketing AI Institute’s newsletter and stay informed on the latest in AI.