Welcome to #50

In this issue:

The Pope’s Warning: AI Could Repeat History’s Worst Mistakes

Visa and Mastercard Enter the AI Agent Era

Is AI Helping or Killing Sales?

Plus, we discuss British Airways huge $7B upgrade, what happens when the final conversation with the deceased never has to end, whether Airbnb’s new “everything app” might actually mean “everything”, updates from our Slack Community, and all the news we found interesting this week.

It’s Sunday morning!

Canada just appointed its first AI minister. Next up: AI ambassadors and robot diplomats.

I bet they will all read this newsletter too.

Don’t fall behind.

The DAS Crew - Andy, Beth, Brian, Eran, Jyunmi, and Karl

The Pope’s Warning: AI Could Repeat History’s Worst Mistakes

Newly elected Pope Leo XIV recently issued a stark warning about artificial intelligence, comparing its rapid and disruptive impact on society to the industrial revolution’s dehumanization of workers. In one of his first major addresses, Pope Leo highlighted the need for careful ethical reflection and strong international cooperation to prevent AI from exacerbating existing social inequalities, as unregulated industrialization did in the 19th century.

Drawing parallels to Pope Leo XIII’s historic 1891 encyclical, Rerum Novarum, which advocated for workers' rights amid rampant industrial exploitation, the Pope emphasized the moral responsibility to ensure AI serves human dignity rather than undermines it. He argues that AI, while offering tremendous potential for societal good, risks concentrating power and wealth even more dramatically than past technological shifts, given its unprecedented speed and scope.

The Pope's intervention is not merely symbolic; it carries significant moral weight globally, calling on tech leaders, governments, and the broader society to consider ethical dimensions alongside technological progress. He also implicitly invites the Catholic Church and other faith-based organizations to proactively engage in shaping a more equitable and humane AI-powered future.

WHY IT MATTERS

Historical Parallels: Comparing AI's rapid societal changes to the industrial revolution highlights urgent moral concerns and underscores the need for proactive rather than reactive responses.

Human Dignity at Stake: Pope Leo XIV's message underscores that technological progress should never come at the cost of fundamental human rights and individuals’ dignity.

Moral Authority and Influence: The Pope's prominent global platform gives significant weight to discussions around AI ethics and regulation, influencing conversations well beyond religious circles.

Call to Action for Institutions: Religious and community organizations could play vital roles in AI education, advocacy, and social programs, addressing ethical concerns and helping mitigate inequalities created by AI’s disruptions.

Balancing Progress and Ethics: The Pope's message promotes a vision of AI development firmly aligned with human-centered values, urging stakeholders to prioritize people over profits.

Visa and Mastercard Enter the AI Agent Era

Visa and Mastercard recently announced major moves into the world of AI-driven payments, signaling the start of a future where your AI assistant handles financial transactions autonomously. Visa's new Intelligent Commerce platform, endorsed in partnership with major tech companies like OpenAI, Microsoft, and Anthropic, enables users to authorize AI agents to conduct purchases on their behalf, within preset limits and personalized rules to maintain control over AI agents financial transactions.

Similarly, Mastercard has introduced Agent Pay, emphasizing the same capabilities for autonomous agents to transact securely, quickly, and efficiently. These developments mean you could soon rely on an AI assistant to find and purchase concert tickets, automatically reorder groceries, or even manage personal or business expenses without manual action and oversight.

However, these innovations come with critical questions.

• How will businesses market directly to AI agents?

• What will consumer trust look like when AI controls spending?

• And importantly, how secure will these transactions be, especially against fraud, unauthorized purchasing, or unintended actions?

Both companies are working hard to ensure their platforms provide robust safeguards, emphasizing security, convenience, and transparency.

WHY IT MATTERS

New Frontier of E-Commerce: AI-driven payment systems could redefine online shopping, fundamentally changing how businesses market and sell products. Companies must learn how to position themselves effectively in an AI-first world.

Consumer Trust is Critical: Successful adoption hinges on consumer confidence in security and transparency. Clear rules, strict limits, and strong fraud detection will be essential to earning user trust.

Competitive Pressure: Visa and Mastercard's announcements set the stage for other payment providers to rapidly follow suit, creating intense competition and innovation in autonomous payment services.

Impact on Local and Small Businesses: Autonomous agents might boost local economies by enabling easy specification of users preference for local purchases, assuming small businesses can integrate effectively with these new systems.

Potential Economic Disruption: Automated purchasing could reshape retail and wholesale markets by shifting spending patterns significantly. Businesses must prepare for these potentially disruptive changes in agent-assisted consumer behavior.

Is AI Helping or Killing Sales?

Sales teams have always aimed for personalization, but AI is pushing this to new extremes through hyper-personalization. Imagine walking down the street and seeing a customized ad displayed on your glasses or windshield, an ad uniquely created just for you, based on your past purchases, current interests, and even your exact location and context. While it sounds futuristic, this level of personalization is becoming a reality, raising significant questions about trust, privacy, and the human’s role in sales.

On one side, businesses see hyper-personalization as a way to provide precisely the right solutions at exactly the right time. AI-driven tools now generate highly customized emails, live-call prompts, and sales materials almost instantly, saving hours of preparation and research. Companies like Skaled are already AI-enabling sales teams to deeply personalize outreach to prospects, significantly boosting engagement and effectiveness.

Yet, there's a significant risk: hyperpersonalization can feel intrusive, manipulative, or just plain creepy to consumers. Trust becomes paramount, especially as AI technology increasingly moves from our screens into our daily interactions and environments. Buyers, too, are starting to use their own AI agents to filter out intrusive marketing messages, creating an escalating battle of "agent versus agent" to control the sales experience.

WHY IT MATTERS

Balancing Personalization and Privacy: Companies must carefully balance delivering highly relevant experiences without crossing into invasive or manipulative interactions.

Reinventing Sales Roles: Sales teams could shift dramatically, with AI handling initial outreach and prospecting, allowing salespeople to focus on high-value human interpersonal interactions.

Agents on Both Sides: As buyers adopt AI to manage incoming sales pitches, sellers need new strategies to engage with AI intermediaries effectively and ethically.

Trust as a Competitive Advantage: Businesses that clearly and transparently manage data and privacy will differentiate themselves in a hyper-personalized market.

Emerging Technologies Drive Change: AR glasses, smart windshields, and other immersive technologies will dramatically alter the nature and strategy of sales and marketing.

Just Jokes

Airbnb expands into services and experiences, plans more social and AI features

Did you know?

British Airways has significantly improved its flight punctuality in early 2025, with 86% of flights now departing on time, thanks to a £7 billion investment in artificial intelligence technology.

The airline achieved more than 90% on-time departures on 38 of the 89 operational days in the first quarter, and two-thirds of April departures from Heathrow left ahead of schedule which is double the 2023 rate.

The AI-driven improvements include real-time weather routing to avoid delays, smart aircraft stand allocation based on live passenger data, and new digital tools for pilots and crew, collectively saving hundreds of thousands of minutes in delays. BA’s ‘Mission Control’ system unifies previously separate operational technologies into a single global interface, enabling rapid, preemptive decisions.

Over 100 data scientists support these innovations, with £100 million specifically allocated to developing digital tools. These advancements have led to smoother travel experiences and more efficient ground operations. BA Chairman Sean Doyle emphasized at an aviation summit that leveraging AI, machine learning, and forecasting tools is transforming the airline's resilience and performance, making air travel more reliable for customers.

Source

Justin shared his vision from Thursday’s show

This image was inspired by our conversations about AI agents in advertising and the invisible algorithms shaping our daily lives.
What a wild time to be alive!
It only took 7 iterations to get the spelling right and fix one missing arm. Probably enough energy used to generate the image to power a small town, so I’ll just leave this one here.

Gareth and Andy talk about Double Vibe Coding

After sharing this image with the community of Gareth working in Lovable and Gemini at the same time, Andy responded:

I’m with you, Lovable isn’t as smart as Gemini 2.5 Pro and doesn’t have the same context retention, so I have been using Gemini as an assistant, reviewing Lovable’s analysis and code change plans, Gemini generates instructions to Lovable that I cut and paste. It’s amazing how much more Gemini understands and corrects, so much so that as soon as I can stabilize the lovable.Dev app I’m going to try to port it over to Firebase Studio.

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

The Resurrection Memory Conundrum

We’ve always visited graves.
We’ve saved voicemails.
We’ve played old home videos just to hear someone laugh again.

But now, the dead talk back.

With today’s AI, it’s already possible to recreate a loved one’s voice from a few minutes of audio. Their face can be rebuilt from photographs. Tomorrow’s models will speak with their rhythm, respond to you with their quirks, even remember things you told them, because you trained them on your own grief.

Soon, it won’t just be a familiar voice on your Echo. It will be a lifelike avatar on your living room screen. They’ll look at you. Smile. Pause the way they used to before saying something that only makes sense if they knew you. And they will know you, because they were built from the data you’ve spent years leaving behind together.

For some, this will be salvation. A final conversation that never has to end.

For others, a haunting that never lets the dead truly rest.

The conundrum
If AI lets us preserve the dead as interactive, intelligent avatars, capable of conversation, comfort, and emotional presence, do we use it to stay close to the people we’ve lost, or do we choose to grieve without illusion, accepting the permanence of death no matter how lonely it feels?

Is talking to a ghost made of code an act of healing? Or a refusal to be human in the one way that matters most?

Want to go deeper on this conundrum?
Listen/watch our AI hosted episode

OpenAI Adds PDF Export for Deep Research Results
OpenAI added a long-requested feature to its deep research tool: the ability to download results as a PDF. The functionality allows users to more easily store or share structured outputs from complex multi-step research queries.

Deeper Insight:
While simple, this feature shows OpenAI’s growing focus on making agentic outputs easier to reuse in professional settings. It also signals a trend toward positioning ChatGPT as a full-stack research and knowledge management platform.

Absolute Zero Reasoner Demonstrates Self-Learning AI Capabilities
A collaboration between Tsinghua University, Penn State, and others produced the Absolute Zero Reasoner, a framework for training reasoning models from scratch without human data. The model teaches itself by generating and solving its own problems and improves through millions of self-play iterations.

Deeper Insight:
This project shows the growing viability of self-learning systems in reasoning tasks. While initially focused on code, the implications for generalized learning, autonomous decision-making, and AGI development are significant.

Sakana AI Introduces Continuous Thought Machine Inspired by Neural Timing
Sakana AI launched a new framework for reasoning that mimics the timing dynamics of biological neurons. The system solves problems using synchronized neuron activity, allowing more interpretable and step-by-step reasoning.

Deeper Insight:
By modeling the temporal behavior of real brains, Sakana may open new doors for explainable AI. Their biologically inspired approach could improve how machines simulate decision-making and make reasoning more transparent to humans.

Saudi Arabia Announces $40 Billion AI Investment Zone with AWS Involvement
Crown Prince Mohammed bin Salman announced a $40 billion AI initiative requiring local data residency for participating companies. Amazon Web Services was named as a key infrastructure partner, with data centers and edge computing to support development.

Deeper Insight:
This announcement reflects the geopolitical shift toward regional AI hubs with strict data localization. Saudi Arabia’s move follows similar pushes in the UAE and China, marking a growing fragmentation of the global AI ecosystem.

Trump Administration Rescinds AI Chip Export Restrictions
The Trump administration reversed Biden-era semiconductor export controls and replaced them with a country-by-country commercial agreement model. Executives from Microsoft, Nvidia, and other U.S. tech firms participated in the policy summit in Saudi Arabia.

Deeper Insight:
This policy shift favors business-friendly diplomacy over broad regulatory guardrails. While it may unlock new markets for U.S. AI hardware, it also opens doors to controversial partnerships and raises questions about long-term national security.

U.S. Commerce Department Declares Global Use of Huawei AI Chips a Violation
In a sweeping move, the Commerce Department announced that any global use of Huawei’s Ascend AI chips could be considered a violation of U.S. export laws, regardless of where the chips were purchased or deployed.

Deeper Insight:
This extraterritorial stance suggests the U.S. is extending its AI export policy beyond domestic enforcement. It sets a potentially dangerous precedent for how governments regulate international tech usage and complicates global AI deployment.

FDA Appoints First Chief AI Officer to Modernize Approvals
The FDA named Jeremy Walsh as its first Chief AI Officer to accelerate medical and device approvals using generative AI. Initial pilots suggest approvals that took days can now be completed in hours.

Deeper Insight:
If successful, this could radically streamline healthcare innovation and increase access to life-saving technologies. But the challenge will be balancing speed with the rigorous safety standards the FDA is known for.

OpenAI Launches HealthBench to Evaluate AI in Clinical Reasoning
OpenAI debuted HealthBench, a new benchmark that tests AI models on real-world clinical decision-making. Built in partnership with 262 physicians who have practiced in 60 countries, HealthBench includes 5,000 realistic health conversations, each with a custom physician-created rubric to grade model responses. OpenAI’s o3 model performed best on the questions, ahead of Grok 3 and Gemini 2.5 Pro. When expert doctors collaborated with AI to take the test, in some of the test areas they were better than the model alone. But in some test areas, AI outputs alone could not be improved by human experts , meaning the AI had responses that were as good or better than human experts alone.

Deeper Insight:
HealthBench adds weight to the argument that advanced models are ready to support real-world medical practice. It also raises ethical and regulatory questions about how much autonomy AI should have in high-stakes environments like medicine.

FDA Approves Precision Neuroscience Brain Implant for 30-Day Trials
Precision Neuroscience received FDA clearance to test a brain-computer interface implant thin enough to be inserted through a one-millimeter opening. The device could help people with paralysis control external devices using thought alone.

Deeper Insight:
This could be a milestone for neurotechnology accessibility. It also signals a move from long-term brain-computer interface research into practical, near-term therapeutic applications.

Apple Collaborates with Synchron on Brain-Controlled Device Interfaces
Apple is working with neurotech firm Synchron to develop standards for brain-computer interface control of iPhones, iPads, and Vision Pro headsets. The Synchron product is a stent-like sensor inserted via catheter, placed in a small vein near the motor cortex.

Deeper Insight:
Apple’s move into neural interfaces could redefine human-computer interaction. If successful, it may set a new standard for inclusive design and open up new possibilities for communication, productivity, and mobility.

MIT Builds Robotic Walker With Airbag to Prevent Elder Falls
MIT engineers introduced Ibar, a robotic walker designed to assist the elderly in standing, sitting, and walking. It includes an automatic airbag system to reduce injury in case of falls.

Deeper Insight:
Ibar shows how robotics can serve aging populations with mobility challenges. As eldercare becomes a critical global issue, human-safe robots like this could play a central role in care systems.

Tesla Demonstrates Optimus Robot at Saudi Summit
Tesla showcased its Optimus humanoid robot dancing and performing simple tasks at the recent Saudi tech summit. While still tethered and not fully autonomous, the robot displayed enhanced hand dexterity and motion smoothness.

Deeper Insight:
Tesla’s progress is more about signaling intent than present-day functionality. With a $30,000 target price, the company is betting on widespread household adoption, even as rivals like Unitree aim to undercut it at half the cost.

EPFL Develops Robotic Hand With Emergent Grip Behavior
Researchers at EPFL developed a robotic hand that can pick up 24 different objects using spontaneously emerging movement. Rather than pre-programmed grips, the hand’s material design enables human-like adaptability.

Deeper Insight:
By shifting focus from programming to physical structure, this project reflects a broader trend in embodied AI. Designing intelligence into hardware may be as important as advancing software reasoning.

Caltech Engineers Build Jellyfish-Inspired Ocean Robots
To combat turbulent undersea conditions, Caltech developed autonomous underwater vehicles that mimic jellyfish movement and ride ocean currents for data collection. The robots aim to improve exploration and environmental monitoring.

Deeper Insight:
This bio-inspired approach highlights how AI-driven robotics can learn from nature to operate efficiently in extreme environments. It could expand our ability to monitor climate change and marine ecosystems.

Trump Fires U.S. Copyright Office Director Following AI Fair Use Report
After the Copyright Office questioned whether AI companies can claim fair use when scraping copyrighted material, the Trump administration abruptly dismissed its director. The office had advocated for licensing frameworks instead of broad fair use interpretations.

Deeper Insight:
This action signals mounting political interference in AI-related regulation. It may further entrench industry power at the expense of public oversight, especially as copyright lawsuits against AI firms heat up.

UK Government Pilots AI Tool to Process Public Feedback
The UK tested an AI system called Consult to process public consultation responses. The tool matched human performance in accuracy and nuance, and could save 750,000 hours annually.

Deeper Insight:
If scaled, this could transform how governments engage citizens. But it also raises concerns about transparency and how AI shapes policy input behind the scenes.

Google to Embed Gemini 2.5 Pro Across Android Ecosystem
Google announced that Gemini 2.5 Pro will become the default AI layer across Android, powering everything from smartwatches to cars. It will also support conversational driving in Android Auto.

Deeper Insight:
This integration gives Android a unified AI identity, posing a major challenge to Apple, which has yet to announce a comprehensive device-wide AI strategy.

NVIDIA Releases Open-Source Code Reasoning Models
NVIDIA open-sourced three code reasoning models ranging from 7 to 32 billion parameters. These models outperform OpenAI’s o3 mini on the LiveCode benchmark and are optimized for running on local devices.

Deeper Insight:
NVIDIA continues to move beyond hardware with powerful open-source AI tools. By releasing small, high-performing models, it supports decentralized development and democratizes access to advanced coding agents.

Manus AI Agent Platform Now Free to Use
Manus, an agentic platform for complex multi-step tasks, removed its waitlist and is now free to use. Users receive one task per day and a 1,000 credit bonus.

Deeper Insight:
This move opens the door for broader experimentation with autonomous agents. As more users try tools like Manus, the bar for consumer-ready agentic AI will rise quickly.

Aalto University Study: People Judge AI Creativity Based on Process Visibility
A new study found that people are more likely to view AI as creative if they can see how it arrived at its result. Process transparency influenced whether AI-generated drawings were perceived as original.

Deeper Insight:
This research could influence how AI tools are designed, especially in creative industries. Visible workflows may enhance user trust and affect legal judgments around authorship and originality.

TikTok Launches 'AI Alive' Image-to-Video Tool
TikTok introduced a feature that animates still images into short videos based on user prompts. It’s part of the platform’s ongoing push to integrate generative media tools.

Deeper Insight:
By lowering the barrier to animated content, TikTok may fuel a new wave of meme-based storytelling. This tool could also set a precedent for how entertainment platforms deploy AI natively.

Microsoft’s WizardLM Team Reportedly Moves to Tencent
The core researchers behind WizardLM, a Beijing-based team affiliated with Microsoft, are now working at Tencent. This move could significantly shift AI talent and model innovation in China.

Deeper Insight:
Talent shifts in AI have outsized impacts due to the small size of top research teams. If confirmed, this move strengthens Tencent’s position and suggests rising competition among China’s leading AI labs.

Legal AI Platform Harvey Expands Beyond OpenAI
Harvey, the legal AI tool funded by OpenAI, now uses models from Anthropic and Google as well. The company no longer relies solely on OpenAI’s technology.

Deeper Insight:
This diversification reflects growing demand for model flexibility and reliability. As AI tools serve regulated industries like law, companies will seek out best-in-class models to provide the “thinking” for their applications. Separate selection of the core “reasoner” model orchestrating the solution with fine-tuned expert models allows for testing and evaluating for the best combination.

Carnegie Mellon Creates LegoGPT to Generate Buildable Brick Designs
CMU researchers introduced LegoGPT, a model that generates stable and buildable Lego structures based on text prompts. The goal is to combine creativity with engineering feasibility.

Deeper Insight:
LegoGPT merges user-prompted generative AI with real-world product and physical constraints, highlighting how digital tools can contribute to imaginative building and hands-on learning. It’s an interesting example of how AI can support the blending of education and play.