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

• Sam Altman’s Warning

• AI Agents Are Coming—But Are They Really What You Think?

• Why Perplexity’s Spaces Might Change How Your Team Collaborates

• How AI Could Reshape Our World—For Better or Worse

Plus, we discuss Claude’s new computer use (and why bathroom breaks might be out), open sourcing with agent.exe, AI that finally gets the Aussie “accent”, Google’s quest to do away with RAG, people protecting bullied AI’s, and of course all the news we found interesting this past week.

It’s Sunday morning.

Time to dive into AI —because one day, you might need to negotiate with your Roomba.

The DAS Crew

Sam Altman’s Warning

Sam Altman, CEO of OpenAI, recently raised a critical concern: Is society prepared for the rapid pace of AI development? The conversation isn’t just about breakthroughs, but about the ethical dilemma of advancing technologies at a speed that risks leaving entire industries and workforces behind. It’s not a question of if disruptions will happen, but how fast and how deeply they’ll reshape the world as we know it.

The acceleration of AI development feels fundamentally different from past technological revolutions. When cars replaced horses or the internet transformed communication, there was time for society to adapt over decades. But AI is advancing in months, not years. This rapid pace means that businesses, workers, and institutions have a shrinking window to adjust and find new roles in a landscape being reshaped by automation, AI agents, and multimodal capabilities.

The real question: Can the institutions, workers, and public at large adapt quickly enough to handle the sweeping disruptions coming their way?

WHY IT MATTERS

A Tsunami of Change: AI advancements are creating ripple effects in almost every industry, from logistics and customer service to financial services and even creative work. The speed at which this is happening leaves little room for strategic adaptation.

Job Displacement Accelerates: Unlike in previous technological shifts where people had years or decades to adjust, this time workers are finding their roles displaced in mere months. The threat of rapid, widespread job displacement is real and immediate.

Uneven Adoption: While some companies are quickly adopting AI and retraining employees, many are lagging behind, unable to keep up with the blistering pace of change. This disparity could lead to greater economic divides.

Ethical Responsibility in AI: Altman’s warning puts a spotlight on the ethical dilemma faced by AI companies. Should they slow down development to allow for societal adaptation, or keep advancing to maintain a competitive edge?

The Stakes Are Global: The fear isn’t just about being left behind technologically; it’s also about the consequences of letting the wrong actors get too far ahead. This adds a layer of urgency to an already complex situation.

AI Agents Are Coming—But Are They Really What You Think?

As AI technology continues to advance, the term “agent” is becoming a buzzword in the industry, but what exactly qualifies as an AI agent? The debate around this term is intensifying as companies like Salesforce, HubSpot, and Microsoft throw it around to describe their new AI-driven workflows. The big question is whether these so-called agents are truly autonomous or simply glorified automations with a sprinkle of reasoning.

There’s a growing push for agents that are not confined to specific platforms but are capable of acting autonomously across various tools and ecosystems. Imagine a future where an AI agent could coordinate tasks across your entire digital workspace—initiating actions in Salesforce, executing workflows in HubSpot, and analyzing data in spreadsheets—without being limited by platform boundaries. This shift would require moving beyond predefined workflows to agents that can interpret goals, learn, adapt, and act with minimal human oversight.

The industry isn’t quite there yet. Most current AI agents are still tethered to specific platforms, functioning more like advanced workflows that rely heavily on predefined triggers. But the ambition is clear: developers and innovators are striving to build agnostic AI agents capable of acting independently across different software, pushing us closer to a world where true digital coworkers could exist.

WHY IT MATTERS

Beyond Buzzwords: Many so-called agents are just complex automations with some basic decision-making abilities. True AI agents will need to demonstrate autonomy, adaptability, and goal orientation.

Platform-Agnostic Agents: The next big leap will be agents that can move fluidly across multiple digital tools and platforms, breaking free from single-ecosystem constraints.

Efficiency Boost: Platform-agnostic agents could revolutionize how businesses operate, streamlining processes across departments and reducing manual effort for complex workflows.

Real Digital Coworkers: The ultimate goal for AI agents is to become digital coworkers that can independently take on tasks, reason through challenges, and collaborate with both humans and other AI agents.

Preparation is Key: Businesses should start refining their processes today, breaking them down into smaller tasks that could eventually be delegated to more advanced AI agents as they emerge.

Why Perplexity’s Spaces Might Change How Your Team Collaborates

Perplexity recently unveiled a suite of new tools aimed at redefining how users search, share, and collaborate with knowledge. Their latest updates include “Spaces” and an enhanced internal search feature, marking a clear move towards enterprise-level solutions. Spaces allows users to create collaborative environments where internal documents can be seamlessly integrated with Perplexity’s AI-powered web search. This means employees can not only access company-specific data but also supplement it with up-to-date information from the web, all in one place.

What sets Perplexity apart is its combination of proprietary search and retrieval-augmented generation (RAG) features, which allow organizations to centralize knowledge while still tapping into external sources for broader context. Think of it as a high-powered, integrated answer engine, not just a traditional search engine.

WHY IT MATTERS

Smarter Knowledge Sharing: Spaces offer a flexible way to organize and share knowledge within teams, bridging gaps between departments or even entire organizations.

Seamless Integration: By combining internal data with real-time web information, Perplexity reduces the friction of jumping between sources, saving valuable time for employees.

More Than a Search Engine: Perplexity’s proprietary search, combined with its collaborative features, positions it as an answer engine that can effectively deliver contextual and accurate results for complex queries.

Competing with the Giants: In a market dominated by Google and Microsoft, Perplexity’s nimbleness and rapid updates give it an edge in serving enterprises looking for clean, customizable search solutions.

Curation as a Business Strategy: As more businesses and individuals grapple with information overload, Perplexity’s tools offer a way to curate and refine knowledge for specific needs, whether in education, business, or research.

Amodei’s Vision: How AI Could Reshape Our World—For Better or Worse

Dario Amodei, CEO of Anthropic and former VP at OpenAI, recently outlined a detailed vision of AI’s future in his essay, Machines of Loving Grace. While his approach is deeply optimistic, highlighting AI’s potential to solve major global challenges, Amodei doesn’t shy away from addressing the risks. His essay paints a future where AI systems aren’t just powerful but capable of transformative change in key areas like biology, mental health, economic development, governance, and even the meaning of human work.

Amodei’s view centers on the concept of “powerful AI”—machines that surpass human intelligence in nearly every domain, capable of solving the world’s most complex challenges and operating at a scale akin to “a country of geniuses in a data center.”

But with that promise comes a profound ethical dilemma:
How do we harness such power responsibly without creating even greater divides or new risks for society?

WHY IT MATTERS

Redefining Human Longevity: AI could accelerate progress in life sciences, potentially extending the human lifespan to 150 years. This would necessitate significant policy and societal changes to avoid overwhelming resources and reshaping generational dynamics.

Economic Inequality Concerns: While AI could help uplift developing countries, Amodei is cautious about whether it can truly close the gap between rich and poor or address inherent inequalities driven by corruption and concentration of power.

AI in Governance: There’s potential for AI to enhance legal systems and promote fairness, but it could also empower authoritarian regimes if misused. The stakes are high when machines are intertwined with power structures.

A New Era of Work: Amodei anticipates that AI will gradually surpass human contributions to GDP, raising questions about the future role of work in a society where economic productivity is increasingly driven by machines.

Kindness and Cooperation: At the heart of Amodei’s vision is the idea that advancing AI requires grounding our development in core human values like kindness and fairness—a challenge for leaders shaping this new reality.

Just Jokes

Of Course There Is An Open Source Version

Karl shared how Claude’s computer actions already had an open source version. Kyle Corbitt shared agent.exe which he says is free open-source Mac/Windows/Linux app that lets you use Claude 3.5 Sonnet to control your computer.

You can read the full X post here.

ChatGPT Might Finally Understand A Man From Perth

Eran shared he had a much improved experience using ChatGPT advanced voice mode.

Here’s what he said:
It seems my first attempt when I got access to it was not the newest greatest version. Tonight's chat was fast & responsive, albeit with some cut outs on occasions.

What was perhaps the most interesting to me was the fact that the transcript it creates in the background is SUPER-ACCURATE with my voice.

Much better than how Descript works with my aussie accent.

Google Wants to Make RAG Unnecessary

Beth shared info about Google’s Gemini Long Context competition. Google says, “A differentiating factor for the Gemini 1.5 model is its large context window that supports context caching.

This competition is an open-ended call-to-action to share public Kaggle Notebooks and YouTube Videos demonstrating interesting use cases for Gemini 1.5's long context window.”

Want to take a shot at the top 4 prizes of $25k? You can learn more here.

Did you know?

A recent study by Imperial College London revealed that humans show empathy towards AI bots who are left out in social interactions. In a virtual game called Cyberball, participants tended to favor AI bots that were excluded from play, throwing the ball to them more often to compensate for the perceived unfairness. Interestingly, older participants showed a stronger inclination to protect the AI bots, reflecting a human tendency to treat even virtual agents as social beings.

These findings suggest that our interactions with AI aren’t purely transactional—our social instincts kick in, even when we know we're dealing with virtual agents!

Source: Imperial College

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

The AI Free Will Paradox:

As AI becomes more integrated into everyday decisions, from choosing our entertainment to curating our social media feeds, it subtly influences our preferences and choices. AI algorithms are designed to optimize engagement and satisfaction, but by doing so, they may be shaping what we desire or believe, without us even realizing it.

While some argue that AI is simply reflecting and amplifying our own choices, others worry that it’s quietly eroding our sense of free will, nudging us toward decisions we didn’t consciously make.

The conundrum: Is AI enhancing our freedom by offering personalized experiences tailored to our preferences, or is it undermining our autonomy by subtly steering our choices and behaviors in ways we don’t fully understand?

Anthropic Unveils Claude 3.5 and Expands API Capabilities

Anthropic released Claude 3.5, showcasing significant improvements over previous versions. The new model outperforms Opus while also providing limited computer control capabilities. This computer control feature lets users interact with applications like Replit, Asana, and DoorDash, allowing Claude to take real-time screenshots and guide users or execute tasks. The introduction of computer use hints at the potential for more advanced integrations in the future.

Runway's Act-One Tool Offers Expressive Character Animations

Runway launched its Act-One tool, designed to capture and replicate minute facial expressions and movements for character animations. This new technology enables users to generate expressive character performances using simple facial recordings, pushing the boundaries of animation and video production. The model allows one actor to play multiple roles, seamlessly creating complex scenes.

Meta’s MovieGen AI Powers Short Films

Meta introduced MovieGen, its new AI-powered video model. This tool enables filmmakers to generate realistic and creative backgrounds and environments, making it easier to bring their visions to life. In one showcase, a short film titled I H8 AI demonstrated MovieGen’s capabilities, blending old home video footage with new visual elements seamlessly.

AI Model Slivit Accelerates Medical Scan Analysis

A new model named Slivit has been developed to read complex 3D medical scans, specifically MRIs, thousands of times faster than traditional methods. By using advanced vision transformers, Slivit can handle volumetric images, offering healthcare professionals quicker and more detailed insights from 3D scans.

Eleven Labs Introduces Voice Design with Text Prompts

Eleven Labs launched a new feature called Voice Design, allowing users to create custom voices using text descriptions. This tool expands creative possibilities by enabling users to design voices for various characters, making it easier to produce dynamic audio content without needing extensive recording resources.

Crew AI Secures $18 Million to Develop Multi-Agent Platform

Crew AI, a platform for managing and deploying multi-agent AI systems, has raised $18 million at a $100 million valuation. The funding will go toward improving tools for deploying, tracking, and managing AI agents, with a focus on performance measurement and ROI.

Ideogram Launches Canvas for Digital Collaboration

Ideogram unveiled its new tool, Canvas, which provides users with a digital workspace for organizing and managing multiple images and designs. This feature allows for real-time collaboration, making it easier for marketing teams and agencies to work together on creative projects.

2024 State of AI Report Highlights Three Key Trends

The latest State of AI Report identified three major trends in the AI industry: the convergence of top-performing frontier models, increased focus on reasoning and planning in large language models, and the expansion of foundation models into multimodal domains. These trends reflect the industry’s ongoing advancements and shifting focus toward more comprehensive AI capabilities.

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.