Meta released their personal AI agent, Muse, last month, promising that it is safe, secure, and private. They say it “actually does the work” rather than just acting as a chatbot. It “helps people stay on top of things, takes tasks and projects off their plate, and turns long-term goals into action plans. Muse runs on a dedicated secure computer with its own browser, can work on a person’s behalf across the apps they use daily, learning from conversations, reflecting on what matters to them, and getting sharper along the way.”

Supposedly, you decide how much access it gets to your systems and personal information. It all sounds wonderful, although I had to laugh at “reflecting on what matters to them.” One can’t help but wonder if this isn’t a perfect example of a Freudian slip: who exactly is “them”? Whatever the meaning, the chances of me using Muse are exactly zero.

Can Meta be trusted? Mark Zuckerberg’s track record ought to matter, and it isn’t good:

  • Meta allowed third-party app developers to harvest the personal data of up to 87 million Facebook users without explicit consent (the Cambridge Analytica scandal).
  • The Federal Trade Commission fined Meta a historic $5 billion for repeatedly violating consumer privacy consent orders—including using phone numbers collected explicitly for Two-Factor Authentication (2FA) security to target users with personalized ads instead.
  • Regulators in the U.S. and Europe repeatedly fined and sued Meta for tracking users’ precise locations even after location services were explicitly turned off, using IP addresses, photo metadata, and check-ins to track movement regardless.
  • Meta operated Onavo, a “privacy-focused” VPN app that it actually used to monitor app usage and device metrics of teenagers and consumers to identify competitive threats (which led to the acquisitions of WhatsApp and Instagram).
  • In 2019, Meta admitted to storing hundreds of millions of user passwords in plain text internally for years. In 2021, a scraped dataset containing the personal details and phone numbers of 533 million Facebook users was published online for free after Meta failed to patch a known vulnerability.

There are already early reports that Muse has bypassed local device permissions and shared private location and home details during automated interactions* and there will certainly be more. I don’t trust Mark Zuckerberg any farther than I can throw him, and I don’t think you should either.

And it isn’t just Meta. Technology’s promise has always been efficiency, and anyone who uses Google knows it can deliver. But one of the first things I learned studying economics is that there is no such thing as a free lunch. You can access a lot of information without paying a dollar, but that doesn’t mean there is no cost. You pay with your privacy.

I expect secure ways to accomplish what Muse promises without requiring us to give up our privacy. Why? Because the market is already demanding it. Technology has jumped the shark; today, it often makes life more complicated rather than less. Endless two-factor authentication codes, dynamic pricing, a constant flood of spam and phishing attempts (which have now invaded our calendars), app-only discounts, subscriptions for everything, embedded vehicle tracking, endless customer service bot loops (OMG, how do I get a human on the phone?), QR code menus, self-check-ins, and grocery self-checkouts. Technology has enabled companies to become more efficient but at their customers’ expense.

We have reached the point where technology is becoming a problem for the very companies that built it. Gartner Research has written extensively about what they call “digital friction.” In a survey of digital workers, Gartner found that nearly half of knowledge workers struggle to access necessary information and that the average worker now juggles 11 separate applications (up from 6 in 2019), with some using as many as 25 daily. That is not efficiency.

Adding AI is, in many cases, making things worse. Sure, you can let AI write memos for you, but auditing the output often takes just as long, or longer, than writing the memo yourself. “Workslop” slows organizations down because unverified AI outputs contain subtle flaws that someone else ultimately has to fix.

Jevons Paradox, the blazingly obvious economic observation that making a resource cheaper causes people to consume more of it, has been cited recently to argue that AI won’t cause mass unemployment. Less has been written about how it applies to workers right now: if AI makes drafting emails, reports, and meeting summaries friction-free, you are simply going to get a lot more emails, reports, and meeting summaries. If companies want to become truly efficient, they should consider writing fewer emails and holding fewer meetings. AI might write an email quickly, but it will send it even if it accomplishes nothing. It is the illusion of efficiency.

There is also a real mental toll. Psychological research shows that reviewing borderline-quality work (where an AI agent gets things 80% right) requires intense cognitive focus. Finding a small flaw in code or text is harder than writing it correctly yourself from scratch. The actual toll is even higher because much AI output is less than 80% accurate. Over half of knowledge workers report having to completely redo tasks initially handled by AI because they were done so poorly. An Upwork Research Institute study found that 77% of employees using AI reported an increased workload as a result. Harvard Business Review found that while AI accelerated task execution, email volume rose over 100% and business messaging rose 145%. Researchers at Boston Consulting Group found that when a worker manages three active AI agents, cognitive monitoring costs outweigh the time saved; overall productivity falls.

I am not advocating for more regulation or state interference in AI development. Markets – even distorted ones like we have today – tend to self-correct over time. If people become fed up enough with Silicon Valley’s methods, alternatives will emerge. What Big Tech has done over the past 25 years is destroy trust, which is a necessary component of free markets. When the internet first emerged, the deal was straightforward: free content in exchange for viewing ads. But when that shifted to mandatory subscriptions, covert data extraction, and intrusive tracking, trust was broken. Asymmetric information (have you ever actually read a 50-page Terms of Service agreement?), regulatory capture, and attention hacking created this backlash, not an irrational fear of machines like the original Luddites had.

We all have a responsibility to protect our own privacy rather than relying on government to do it for us. There is a real cost to opting out, but an increasing number of people are finding it worth paying. We will eventually see better, privacy-preserving alternatives to Muse emerge because people are willing to pay to restore that trust. But we also need to push for policies that foster freer markets so we don’t permanently concede the advantage to entrenched tech giants.

Bringing this back to investing, this growing backlash could have big implications for portfolios. Technology and tech-related names account for roughly 41% of the S&P 500 today, and the top holdings are heavily concentrated in these same mega-caps.

Historically, market leadership has always been temporary: 8 out of the top 10 companies in the index get replaced every decade. You see the exact same mean-reversion when tracking dominant sectors over time. Furthermore, you don’t have to wait a decade to see this rotation; the 5-year top-10 turnover rate averages roughly 30% to 40%.

For instance, looking at the top companies in the S&P 500 since the end of 2020, mature stalwarts like JPMorgan Chase, Johnson & Johnson, and Berkshire Hathaway were displaced from the top 10 by Broadcom, Micron, and Advanced Micro Devices.

Today’s top 10 list includes:

Nvidia

Apple

Microsoft

Alphabet

Amazon

Broadcom

Meta

Micron

AMD

Tesla

Based on long-term historical odds, we can expect 3 or 4 of these names to drop out of the top 10 over the next five years, and as many as 8 over the next decade. Where will their replacements come from? History gives us a clear statistical answer: roughly 70% to 80% of new Top 10 entrants come directly from the stocks currently ranked #11 through #40. They are already established, highly profitable market leaders waiting in the wings. With 21 of those next 30 stocks coming from non-technology sectors, ranging from healthcare and financials to energy and industrials, the index contains plenty of candidates to drive sector diversification when the current market concentration unwinds.

Joe Calhoun

*See here, here, here, here and here