AI Wealth Truth (47): Why Recommendation Algorithms Make the Poor Poorer and the Rich Richer

I. The logic of algorithmic recommendation is simple: show you more of what you like to watch. Whatever you click, it recommends more of it. It sounds like personalization. In reality, it accelerates class lock-in.

II. The problem is: different classes "like" different content.

III. Data shows: lower-income users consume more entertainment, comedy, relationships, gossip. higher-income users consume more business, investing, technology, and industry analysis. Algorithms reinforce this difference, not bridge it.

IV. If you often watch comedy clips, the algorithm shows you more comedy clips. If you often watch investment analysis, the algorithm shows you more investment analysis. Your information intake gets locked into your current preferences. This is called a filter bubble.

V. What does a filter bubble do to wealth?

VI. Impact 1: the cognition gap expands. Some people consume business-model analysis, macro trends, and investment strategy every day. Others consume celebrity gossip, emotional stories, and funny clips every day. After a year, their cognitive frameworks diverge dramatically. What you consume determines what opportunities you can see.

VII. Impact 2: impulsive consumption is stimulated. Algorithms know who is more likely to buy impulsively. They push more product recommendations, promotions, and shopping content to those users. They push other content to users less prone to impulse buys. Algorithms harvest the most harvestable people.

VIII. Impact 3: time allocation is manipulated. Entertainment content is designed to be more addictive. Scrolling comedy feels much better than reading investment analysis. Algorithms keep feeding you what feels best. You spend more and more time on low-value content.

IX. What is more frightening is: filter bubbles are self-reinforcing.

X. You watch entertainment -> the algorithm shows more entertainment -> you spend more time on entertainment -> you have less time to touch other content -> your preference tilts further toward entertainment... This is a positive feedback loop. You get trapped deeper and deeper.

XI. What does this mean for class mobility?

XII. In the past, the poor and the rich at least read the same newspapers and watched the same TV. Information distribution was broadcast-style. Everyone received the same content. Now, each person sees completely different content. The rich receive information that makes them richer. The poor receive information that makes them poorer.

XIII. This is not a conspiracy. It is the natural result of optimizing for "engagement". It gives you what you like. The problem is: what you like is not necessarily what benefits you.

XIV. In the AI era, this split becomes more extreme.

XV. AI can generate infinite content. Content is customized to each person's preferences. Your filter bubble becomes thicker and less penetrable. You see only what the algorithm thinks you should see.

XVI. AI also makes content more attractive. Funnier jokes, more emotional stories, more precise emotional triggers. Low-quality content becomes more "watchable", making it harder to escape.

XVII. Meanwhile, truly valuable information becomes harder to find. Because it is not recommended to people who "do not like" it. Valuable information requires active search, not passive feeds.

XVIII. How do you break the filter bubble?

XIX. 1. Search actively instead of scrolling passively. Decide what you want to know, then search for it. Do not wait for the algorithm to deliver it. Active information intake is more valuable than passive intake.

XX. 2. Reset recommendation history regularly. Some platforms let you reset recommendations. Or use a new account and retrain the algorithm. Break the existing information loop.

XXI. 3. Deliberately consume different types of content. If you always watch entertainment, force yourself to watch some business content. Even if it feels less satisfying, keep doing it. Expand your information boundary.

XXII. 4. Use non-algorithmic sources. RSS, email newsletters, books. These are not pushed by algorithms. You choose them. Use more pull, rely less on push.

XXIII. 5. Pay for high-quality information. Free content is ad-driven and optimized for clicks. Paid content is subscription-driven and optimized for value. Use money to filter out low-quality information.

XXIV. Recommendation algorithms are not neutral technology. They reinforce your existing preferences and lock you into your current class. If you want to move classes, you need to break your filter bubble. What you see determines what you can become. In the AI era, the walls of the filter bubble get higher and thicker. Breaking through requires more initiative and effort. But the reward of breaking through is also larger.