The hosts discussed how VistaVision offers several practical advantages:
It uses 35mm film stock, making it more economical than IMAX's 70mm format
The format provides a distinct "OG Hollywood vibe" in the final result
It was used by cinematic masters like Alfred Hitchcock
This trend toward traditional film formats occurs alongside advances in digital filmmaking, with the hosts noting that filmmakers like Zack Snyder primarily use RED digital cameras for their ability to capture high frame rate action sequences.
"I'm super happy that all of this is making a comeback with the added benefit of modern VFX, which makes this so magical," Addy remarked.
The hosts emphasized that modern filmmaking allows for multiple approaches, with Joey noting: "It kind of shows that there could be multiple lanes. You could have these big temple blockbuster films and throw everything at it and bring in these cool celluloid, thematic extravaganzas. And then also, we have digital so we could make a lot more stuff and stories and films. You can have both."
Companies Reassessing AI Strategies After Disappointing Results
Shifting to technology news, Joey and Addy discussed recent headlines about companies reconsidering their AI implementations after failing to see expected returns on investment.
They highlighted two specific examples:
Klarna, the buy-now-pay-later service, had previously announced becoming an AI-first company but is now rolling back some AI initiatives and returning to human workers
An IBM study found that only 25% of CEOs report that their AI initiatives have delivered expected ROI, and only 16% have scaled AI enterprise-wide
The hosts provided context for these developments, noting that many companies jumped into AI implementation too early, before the technology was mature enough for certain applications.
"If you were trying to jump into this two years ago, it was pretty crappy," Joey observed. "ChatGPT-3 was kind of cool-ish, but it still needed a lot of hand-holding."
Addy characterized this as "the AI bubble contracting" but emphasized that current AI capabilities are significantly more advanced than those available a few years ago: "If you repeat that same experiment today with today's capabilities, I bet the results would be very different."
The hosts also discussed how AI adoption has been more successful in specific domains, particularly coding and software development, where AI-powered IDEs and development tools have proven highly effective.
"If you look at the big trends of companies like Meta and Google, if you look at the AI use cases across the board, the number one use case seems to be coding and software development," Addy noted. "That stuff is so good."
Joey agreed but expressed concern about relying too heavily on AI without understanding the underlying principles: "The scary thing is if this is all you know, or how all you learn on, what's the future gonna be like when you don't know how to troubleshoot or how this is actually working under the hood?"
Microsoft Research's AI Model That Flags Its Own Inaccuracies
In the final segment, Addy introduced Microsoft Research's new AI model called Claimify, designed to address a common problem with large language models: their tendency to provide incorrect information with high confidence.
"We use ChatGPT every day in our lives... Does it ever come back to you with 'I'm sorry, I don't know that one?" Addy asked, to which Joey confirmed that AI models are often "confidently incorrect."
Claimify addresses this issue through several mechanisms:
Disambiguation: When answers are vague, it forces the model to provide more precise information
Decomposition: It breaks down complex statements and verifies each component independently
Context awareness: It asks for additional context from users to improve answer accuracy
Looking at the research paper, Joey described it as a "high-end fact checker" that analyzes statements, flags issues, and verifies information.
"It's really pertinent in the world where we live with a lot of fake news and just a lot of inaccurate information or AI output," Addy observed. "Who knows if it's true or not?"
The hosts discussed how this technology could be particularly valuable for applications like search engines, news organizations, and academic research, where accuracy is crucial.
"I think this is the most interesting time because now we are seeing this stuff really come into fruition in a useful way," Addy said, noting that we're seeing a new level of refinement in AI applications.
Joey added that different AI models are increasingly specializing in particular tasks: "I think we're getting where all the large models are pretty good baseline right now, and it's figuring out the specialized tools and applications for them."
Conclusion
The diverse topics covered in this episode of Denoised highlight the ongoing dialogue between established filmmaking techniques and cutting-edge technology. Whether it's Christopher Nolan pushing the boundaries of what's possible with IMAX or Microsoft developing AI systems that can verify their own accuracy, the media and entertainment industry continues to evolve through both preservation of craft and technological innovation.
What's particularly notable is how traditional film formats are experiencing a renaissance at the same time that AI capabilities are being refined and specialized. Rather than one replacing the other, we're seeing a layered approach where different tools serve different purposes.
The views and opinions expressed in this podcast are the personal views of the hosts and do not necessarily reflect the views or positions of their respective employers or organizations. This show is independently produced by VP Land without the use of any outside company resources, confidential information, or affiliations.
