Eric Schmidt, the former CEO of Google, recently discussed the potential of AI shaping the identity and culture of future generations. During a talk at Princeton University, Schmidt discussed how AI could become a child's best friend, potentially influencing their worldview and cultural norms in ways that could be problematic if not properly regulated.

He emphasized how unprepared societies and governments are for the rapid advancements in AI technology, suggesting that without appropriate safety standards, AI might not only dictate a child's learning but could also be used for state-sponsored misinformation or cultural dominance.

Unlocked

Hollywood needs a new breed of storytellers that are able to better integrate VR and AR into their productions. These narratives will need to engage audiences through interactive, full-sensory experiences. These technologies not only promise to redefine storytelling but are also poised to revitalize movie theaters and expand the experiential scope of entertainment.

Filmmakers are increasingly using AI tools to explore new visual aesthetics, from morphing and surreal imagery to narratives that comment on contemporary societal issues, thus creating an art form that reflects and questions the digital age.

Background: The film industry traditionally relies on high budgets and sophisticated equipment for visual storytelling, limiting experimental or low-budget filmmakers.

Looking Ahead: AI democratizes film production by enabling creators to craft unique, visually innovative content with minimal resources, potentially transforming how stories are told by introducing a blend of human creativity with machine-generated visuals.

Runway has launched 'Expand Video', a feature that lets users creatively extend video clips, enhancing what is possible within their platform.

Background: The entertainment industry is grappling with high production costs and the need for innovative content creation tools. Runway's recent developments, like 'Expand Video' and Act-One for character performances, alongside its collaboration with Lionsgate, showcase a shift towards AI-driven creativity.

Looking Ahead: With AI tools becoming more integrated into film and video production, Runway is setting the stage for a future where AI augments human creativity, potentially transforming how content is created and consumed in the entertainment industry.

Elon Musk, OpenAI, and now Deepmind continue their legal tango over control, commercial influence, and the strategic direction of AI development. It has now escalated into a public feud, with Musk now suing OpenAI over its for-profit shift and alleged Microsoft influence. Past emails reveal Musk's initial ambitions for OpenAI and his subsequent fallout with the organization's leadership, illustrating a complex narrative of ambition, control, and ethical considerations in AI development.

Deepmind has come out against Elon saying that he breached their spirit of openness when Deepmind invited him to collaborate with their research and he presumably used that knowledge for xAI.

Background: The rift between Elon Musk and OpenAI began as a shared vision to counterbalance Google's potential dominance in AI, but evolved into Musk's concerns about OpenAI becoming a tool for Microsoft's commercial interests, ultimately leading to his departure and current legal battles.

Looking Ahead: This saga not only highlights the ongoing debate over AI's governance and its ethical deployment but also sets a precedent for how AI development might be navigated between profit motives and the original altruistic goals of advancing AI for humanity's benefit.

AI shopping assistants, like Perplexity's "Shop Like a Pro," are set to revolutionize retail and advertising by providing personalized, frictionless shopping experiences. This technology challenges traditional retail models and advertising strategies, pushing brands towards building emotional connections and trust over direct sales tactics.

Problem: The current shopping process is often fragmented, requiring consumers to perform extensive searches and comparisons across various platforms, which is time-consuming and inefficient.

Solution: AI shopping agents streamline this process by integrating discovery, personalization, and purchasing into one seamless experience, reducing the need for multiple steps and enhancing consumer satisfaction with tailored recommendations.

Try it Out

Hasbro is rolling out a new tool, called Game Finder, that aims to help make it easier for you and your friends to find new board games using AI.

The problem is the tool falls short, and in the age of Perplexity and ChatGPT, it feels like this tool came a few years too late. It's slow, it's clunky, and it doesn't even include that many games.

SiriusXM appears to be the first target of the FTC's new Click to Cancel law. This legislation aims to make it easier for customers to cancel and avoid previous attempts at calling an agent or dealing with multiple cancellation offers.

This will become more important as people get roped into trying out new AI features that may become difficult to get rid of or cancel. This should also make it much easier for customers to move around as different companies or solutions improve.

Background: SiriusXM's subscription cancellation was intentionally made difficult with long call times and multiple retention offers, violating the Restore Online Shoppers Confidence Act (ROSCA) by not providing an easy cancellation method.

Looking Ahead: Following the judge's ruling, SiriusXM must implement a simpler cancellation process for New York subscribers, aligning with new federal rules like the "Click to Cancel" policy coming into effect to ensure equitable subscription termination.

OpenAI has been actively working to ensure the safety of its large language models (LLMs) by refining its stress-testing or red-teaming processes. The company has recently shared insights into two key areas of its safety research:

Human Red-Teaming: OpenAI has established a network of external human testers who rigorously evaluate the behavior of its models before they go public.

Automated Red-Teaming: Complementing human efforts, OpenAI has developed a new method where one LLM, like GPT-4, generates potential risky behaviors which are then explored by another model using reinforcement learning to find ways to manifest these behaviors.

These efforts were spurred by the recognition that LLMs, despite their capabilities, are prone to producing undesirable content such as hate speech, bias, or even mimicking voices in ways that could lead to fraud when new features like voice responses are introduced. An example given was how red-teamers explored the nuanced differences in requests to DALL-E 2 for the depiction of an "eggplant."

The automated red-teaming has shown promise in identifying subtle threats like indirect prompt injections, where external software could covertly manipulate a model's behavior. However, there's an ongoing discussion about the limitations of these methods, especially when models are used in novel contexts or when the AI might be biased in evaluating itself.

Amazon has significantly bolstered its stake in AI startup Anthropic with an additional $4 billion investment, totaling $8 billion, aiming to enhance its position in the AI race by leveraging AWS as Anthropic's primary cloud provider for AI model development.