The Pope's AI Encyclical, AI's PR Emergency, and the Soaring Cost of Intelligence

The rapid advancement of artificial intelligence continues to spark widespread debate, from the Vatican's pronouncements to the growing public skepticism and the escalating costs of AI implementation. This week, we delve into the Pope's landmark encyclical on AI, the escalating public relations crisis facing the AI industry, and the surprising financial challenges enterprises are encountering with AI adoption.

The Pope's AI Encyclical: A Call for Human-Centric AI

Pope Leo XIV, the first American Pope, has issued his first encyclical, "Magnifica Humanitas" (Magnificent Humanity), a comprehensive document dedicated to artificial intelligence and the preservation of human dignity in the age of AI. Released on the 135th anniversary of Pope Leo XIII's 1891 encyclical "Rerum Novarum," which addressed workers' rights in response to the Industrial Revolution, Leo XIV's message draws a direct parallel, arguing that AI represents a similar transformative upheaval demanding ethical guidance.

The central thesis of the encyclical is that technology is never neutral, as it "takes on the characteristics of those who devise, finance, regulate, and use it." While acknowledging AI's potential as a valuable tool, the Pope warns that it "tends to amplify the power of those who already possess economic resources, expertise, and access to data." Without oversight, he cautions, "those who control AI will impose their own moral vision, which will become the invisible infrastructure of these systems." He emphasizes that "a more moral AI is not enough if that morality is determined by a few."

A key call to action within the encyclical is to "disarm AI." This is interpreted not solely as a military concern but as a broader economic and cognitive arms race driven by the pursuit of geopolitical or commercial dominance. Disarming AI, the Pope explains, means freeing it from this "mentality of armed competition" and preventing it from dominating humanity. It entails opening AI to discussion and debate, making it "human-friendly" and free from monopolistic control.

The Vatican's decision to unveil the encyclical alongside Chris Ola, co-founder of Anthropic, underscores the significance of this initiative. With over 1.3 billion Catholics worldwide, the Pope's words carry substantial influence, shaping beliefs, ethics, and educational practices. The encyclical also extends its reach to Christians globally and "all men and women of goodwill," aiming to foster a broader conversation about AI's impact on society, international diplomacy, and the need for renewed educational alliances to teach "digital sobriety."

The encyclical touches upon several critical issues, including the illusion of AI empathy, the inherent biases in AI systems, the hidden environmental costs, the invisible human labor behind AI training, data colonialism, algorithmic injustice, the concentration of digital power, workforce deskilling, autonomous weapons, and the impact on education and critical thought.

In response to the encyclical, prominent figures in the AI community have offered their perspectives. Chris Ola, in his remarks, highlighted the incentives and constraints within AI labs that can conflict with ethical considerations, emphasizing the need for external critics to ensure safety and responsible development. He also posed three key questions for discernment: the duty to the global poor in the face of potential job displacement, the need for moral imagination regarding human flourishing, and the ongoing mystery surrounding the internal workings of AI models, which exhibit structures mirroring human neuroscience and internal states that functionally resemble emotions.

The Pope's encyclical marks a significant moment, urging a global dialogue that extends beyond the confines of the technology sector and calls for a more human-centric approach to AI development and deployment.

AI's Public Relations Emergency

The AI industry is facing a growing public relations emergency, marked by increasing skepticism and negative sentiment. This concern was amplified by recent events, including college graduates booing commencement speakers who mentioned AI and dismal polling data revealing widespread opposition to data centers.

Alex Canerit, writing in his "Big Technology" Substack, highlighted that AI is now in a "serious public relations emergency," particularly among young adults, a demographic crucial for shaping future loyalties and preferences. This sentiment is exacerbated by messaging from within the tech industry itself. For instance, venture capitalist Mark Andre's comments on the Joe Rogan podcast, praising AI for its lack of frustration, sickness, or HR complaints, were framed as a hard sell to unemployed new graduates.

Adding to the concern, a Wired investigation revealed that the Department of Homeland Security and the FBI are monitoring a new domestic threat category: "anti-tech violent extremism." Assessments warn that the chaos from emerging AI could fuel large-scale protests and anti-tech violence.

The industry's credibility is further strained by a perceived arrogance and disconnect from the realities faced by workers outside of Silicon Valley. While tech leaders often point to growth in software companies or increased hiring of engineers, these represent a small fraction of the global workforce. The real impact on marketing, sales, customer success, and HR teams is often overlooked or downplayed.

In response to this crisis, various AI models were prompted to devise a 10-step plan to combat negative sentiment. Key recommendations included shifting from selling "inevitability" to "agency," focusing on AI's role in empowering workers rather than replacing them, and reframing AI literacy as civic literacy. A recurring theme was the need to replace tech CEOs as primary messengers, given their diminished credibility, and to move from abstract benefits to visible, local proof of AI's positive impact.

The consensus is that the AI industry's PR problem is not merely a messaging issue but a fundamental challenge rooted in the tangible impacts of AI on jobs and communities. Addressing these concerns head-on, rather than glossing over them, is crucial for rebuilding trust and fostering broader societal acceptance.

The Soaring Cost of Intelligence

The rapid adoption of AI within enterprises is leading to a significant and often unexpected surge in costs. Many companies are finding that their AI budgets are being depleted far faster than anticipated, prompting a scramble to manage expenses and re-evaluate their AI strategies.

Reports from outlets like Axios and The Wall Street Journal indicate that some enterprises have exhausted their entire annual AI budget within just three months, while others have seen their spending double or triple. Microsoft reportedly canceled internal cloud code licenses due to cost, and Uber's CTO admitted the company blew through its entire 2026 budget in four months, noting that higher token usage wasn't translating into proportionally more useful features. One company allegedly spent half a billion dollars in a single month due to a failure to implement usage limits on its cloud licenses.

The core of the issue lies in the escalating demand for "tokens," the basic unit of measurement for AI computing. As AI models become more capable and agentic, they require significantly more tokens than traditional chatbot interactions. This insatiable demand, particularly in the early stages of AI adoption, is outpacing companies' ability to budget and manage these costs effectively.

The situation is compounded by the fact that many AI budgets were set before the recent explosion in AI capabilities, making them quickly obsolete. Executives in charge of AI access and token budgets are struggling to find solutions for managing, monitoring, and metering usage. Even for smaller organizations, running into daily limits on standard licenses has become a common frustration.

Goldman Sachs forecasts that AI agents will boost tech cash flow, predicting a 24-fold increase in token consumption by 2030. This surge highlights the massive demand, with Google alone processing trillions of tokens per month.

Strategies to mitigate these costs are emerging, including focusing on planning and execution, using smaller models, improving prompting skills, and developing AI literacy. However, the fundamental challenge remains: the desire for more powerful, smarter models to tackle increasingly complex tasks drives up token consumption. The question arises whether current pricing models, which often meter usage by tokens, are sustainable for broader knowledge work, where the value of AI often lies in unlimited access rather than granular consumption tracking. Many believe that a shift towards outcome-based or flat-fee pricing will be necessary for widespread adoption and financial predictability.

Claude Opus 4.8: A Modest but Tangible Improvement

Anthropic has released Claude Opus 4.8, an upgrade to its Opus 4.7 model. The company describes it as a "modest but tangible improvement," achieving better scores across coding, agentic tasks, reasoning, and knowledge work. The model is available at the same pricing as its predecessor: $5 per million input tokens and $25 per million output tokens.

A notable improvement highlighted by Anthropic is Opus 4.8's enhanced honesty. The company states that while all its models are trained to be honest, Opus 4.8 is more likely to flag uncertainties and less prone to making unsupported claims, addressing a common issue where AI models can confidently present thin evidence as fact.

The announcement also detailed new features, including dynamic workflows in Claude Code, allowing for the planning and execution of hundreds of parallel sub-agents in a single session with output verification. Additionally, a new "effort control" setting in the Claude web app and Co-work enables users to select varying levels of effort (low, medium, high, extra, and max) for Claude's responses, with higher effort levels consuming more tokens.

Anthropic also mentioned that models of "Mythos class" are being developed with stronger cybersecurity safeguards and are expected to be released to all customers in the coming weeks. This suggests a potential future release akin to Claude 5. The effort levels in the web app, while useful for developers building repeatable services, present a challenge for knowledge workers performing diverse tasks daily, as the token consumption remains opaque.

The Narrative Around AI and Jobs: Conflicting Data and Lived Realities

A significant disconnect persists between optimistic projections of AI's impact on jobs and the growing evidence of layoffs and anxieties among workers. While some economists and administration officials argue that AI-related job losses are not yet evident in the data, many business leaders and employees are experiencing a different reality.

Reports from sources like Apollo's chief economist and Yale's Budget Lab suggest that current labor market data shows no clear relationship between AI exposure and changes in employment. These analyses often frame broader job anxiety as "largely speculative for now."

However, this perspective stands in stark contrast to the ongoing wave of AI-driven layoffs at major companies like Meta, Block, Atlassian, and Inuit. Furthermore, anecdotal evidence from business leaders and employees paints a picture of significant disruption. Many report experiencing increased productivity through AI tools, which, in a best-case scenario, frees them for higher-value work. In less fortunate circumstances, this increased productivity could automate tasks to the point of job displacement.

The argument is made that relying solely on current data to dismiss job loss fears is shortsighted. Just as early predictions about AI's writing capabilities were met with skepticism, the current data may not yet reflect the full impact of rapidly advancing AI technologies on the workforce. The reality for many businesses, particularly those outside the Silicon Valley bubble, involves tight margins and a need for growth, making the efficiency gains offered by AI highly attractive and potentially disruptive to existing team structures.

The concern is that a narrative of unmitigated abundance and progress, without acknowledging the potential for displacement and underemployment, does a disservice to individuals and society. Preparing for the possibility of significant disruption, even if it proves to be overly cautious, is seen as a more responsible approach than ignoring the potential negative consequences.

AI and Politics: State-Level Regulation and Progressive Resistance

The intersection of AI and politics is becoming increasingly prominent, with developments at both the state and federal levels. Illinois lawmakers have passed SB 315, a bill that AI safety experts believe could become the strongest AI safety law in the country. This legislation requires frontier AI labs to undergo independent third-party verification of their safety commitments, a step beyond the disclosure and incident reporting requirements in California and New York. Notably, both OpenAI and Anthropic have endorsed the Illinois bill, suggesting a de facto national framework is emerging through state-level action.

In contrast, President Trump abruptly postponed the signing of an executive order on AI safety, expressing concerns that a voluntary federal review system could slow the U.S. in its AI race against China.

Meanwhile, a growing progressive resistance to AI is emerging within the U.S. Democratic Party, led by figures like Bernie Sanders, Alexandria Ocasio-Cortez, and Elizabeth Warren. Their proposals include moratoriums on data center construction and new taxes on AI companies, framing the debate around AI's potential to replace human labor.

These developments highlight the increasing politicization of AI and the growing demand for regulatory oversight. While some see the state-level actions as a necessary step in the absence of federal consensus, others worry about a patchwork of regulations. The debate over AI's impact on jobs and society is clearly becoming a significant factor in political discourse.

Microsoft's Work Trend Index: Agents Empowering Humans, Organizations Lagging

Microsoft's 2026 Work Trend Index report reveals that as AI agents take on more of the execution of work, humans are gaining greater agency to direct tasks and own outcomes. The report, based on trillions of anonymized Microsoft 365 productivity signals and a survey of 20,000 AI-using knowledge workers, indicates a significant increase in agent usage, with a 15-fold year-over-year growth in the Microsoft 365 ecosystem.

Key findings include:

The report underscores that while individual capabilities in leveraging AI are advancing rapidly, organizational readiness is lagging. This gap suggests that companies need to focus on fostering supportive cultures, clear leadership alignment, and appropriate talent practices to fully realize the benefits of AI.

AI Use Case Spotlight: Agentic Cold Outreach and HTML Visualizations

This week's AI use case spotlight highlights two practical applications of AI in our work at Smarter X.

Agentic Cold Outreach Experiment: To explore more AI-native approaches to projects like targeted cold outreach, we experimented with Claude Code. By pointing Claude Code at a web page describing a product, it helped identify the ideal audience, roles, and seniorities. In a demonstration, it even attempted to identify potential prospects and guess their email addresses. The most impactful aspect was collaborating with Claude Code to craft a compelling email and then, by connecting it to an email inbox, automatically generate up to 100 email drafts with personalized details. This process, while requiring human relevance and genuine value in the outreach, demonstrated how AI can significantly streamline tasks that might otherwise be time-consuming and prone to copy-pasting.

Interactive HTML Visualizations: Leveraging Claude's ability to generate HTML output has proven invaluable for creating interactive and visual content. By building upon previous HTML outputs and providing specific filters, classifications, and strategic details, we've been able to enrich prompts and generate V2 HTML outputs that serve as minimum viable products for developers. This approach allows for bringing complex ideas to life visually, which is far more effective than simply explaining them in words to a developer. This method of strategic planning and iterative building with AI is transforming how we conceptualize and develop new features and visualizations.

AI Product and Funding Updates

Key Takeaways