#30 Society and Technology

As of December 16 2025, Meta announced it will begin using conversations conducted through its AI-powered chat assistant for ad targeting. This development will directly open ordinary users’ daily conversations to the advertising market. Everything from your vacation plans and health complaints to a simple chat about boredom will become data points marketable in Meta’s eyes.

Until now, the company gathered data from users’ posts likes and clicks to understand their interests. Conversations however occur in a much more direct unfiltered space. They provide a highly accessible medium for easy and candid sharing. Likes and posts were often socially constructed behaviors performed in front of others. But a one-on-one conversation with a chatbot carries far purer more immediate and intention-driven content. This represents nearly flawless targeting potential for advertisers.

Meta also revealed it will integrate this data into its recommendation algorithms alongside its advertising systems. This means the posts videos and news you see on Facebook and Instagram will now be shaped not only by your past behavior but also by questions you direct to AI. For example if you ask about walking you may see ads for hiking groups route suggestions or walking shoes. While this appears designed to enhance user relevance it also seems to deepen Meta’s invisible surveillance of our personal lives by another layer.

Revenue figures are colossal for Meta. The company earned $46.5 billion in advertising revenue last quarter a 21 percent increase compared to a year ago. The addition of conversation data will undoubtedly accelerate this growth in the advertising market.

The European Union the United Kingdom and South Korea are exempt from this practice thanks to strict privacy laws. In the rest of the world users are directed to tools like Ads Preferences to “control” how and to what extent their data is used. Although Meta claims it will not use “sensitive data”—such as health status political views religious beliefs or sexual orientation—for ad targeting the boundary between what is considered sensitive and what is not remains as ambiguous as ever.

At this point the critical question for users is whether this change will be beneficial or whether it will create an experience that has gone “too far.” Seeing content and ads related to our interests can make platforms more useful but the data mining that extends to our conversations reveals that the boundaries of digital privacy have almost entirely vanished. There is no longer any space where you can say “what I said was just a personal note shared with an AI.”

In fact Meta’s move encapsulates the current trajectory of technology. To improve user experience every possible thing that can be measured collected and processed about users has become the backbone of social media technologies. In a sense Meta is transforming conversations from mere exchanges into raw materials for a new data economy. It reminds us that we live in a world where digital surveillance has become an ordinary part of life. Moreover the risk here is not only surveillance. Personal information gathered from likes and shares once carried a degree of social data meaning. But content generated in chatbot conversations represents a far more sensitive form of engineering.

In short Meta is beginning to treat every conversation you have with AI as a kind of “economic signal.” Meta AI used monthly by over a billion people has transformed from a tool that answers your questions into a data engine that drives the company’s billions in advertising revenue. Although it is claimed that usage occurs with user consent practical implementation often reveals that far more of our data is monitored than consent permits.

OpenAI has introduced a new feature called “Instant Checkout” to ChatGPT opening the door to direct shopping via conversation. Users in the United States can now purchase products recommended during a chat with a single click. When users ask for example “What are the best running shoes under $100” they can select one of the suggested items and complete payment and shipping details on the same screen.

At first glance this step appears highly practical. Searching evaluating and purchasing products without leaving the chat eliminates one of the biggest friction points in mobile shopping. But the issue is not merely convenience. The real concern is the increasing convergence of more and more functions within a single application. While search engines e-commerce sites social media platforms and messaging apps previously served distinct functions these functions are now gradually merging into centralized systems like ChatGPT.

The risks of this convergence are alarming. First diversity is disappearing. Instead of comparing products across different sites users may now be limited to what a single platform offers. Although products are said to be ranked organically it is difficult to guarantee that this ranking will not eventually be shaped by commercial agreements or algorithmic preferences. At some point what is recommended to users may no longer be the most suitable option but the one that generates the most profit for the platform. As a result we may find ourselves buying not what we want but what is suggested to us.

Second transparency is eroding. ChatGPT is currently ad-free but there is no guarantee this will last. In the digital economy the strategy “first hook the user then diversify revenue models” has been repeated many times. Today’s convenient “Instant Checkout” could tomorrow merge with in-stream ads turning the chat experience into a hidden billboard behind a commercial facade.

Third there is the issue of privacy. Shopping behavior is one of the most valuable types of data online because it reveals not only your preferences but also your income level lifestyle and even personal needs. When purchases made through ChatGPT are recorded alongside your conversations your personal profile becomes a more cohesive data set. The concentration of such intense information in a single entity increases political and cultural control power.

Despite all these risks there are attractive opportunities for sellers. For small businesses chat-based sales may open a door to customer segments previously unreachable. Yet this also creates a structure that makes sellers dependent on the platform. While customer relationships order processes and data appear to remain with the seller in reality the initial contact guidance and purchasing behavior are all under ChatGPT’s control. This could increasingly trap sellers within OpenAI’s ecosystem.

From the user’s perspective today’s landscape offers the appealing convenience of “find it in chat buy it in chat.” But in the long term this may mean a narrowing of free navigation on the internet habits being directed solely through a single AI and the invisibility of diverse information sources. Users may begin to experience the world through nothing more than the options permitted by “a single door” without even realizing it.

“Instant Checkout” marks a critical turning point in how the internet operates. If everything converges into a single application we users may gain comfort but lose freedom. The critical question that must be asked is this: Does this system that makes shopping easier also narrow our choices? And one day will what we see in ChatGPT reflect not the entire internet but only the fragment a single company chooses to show us?

Liam Fedus’s new initiative Periodic Labs entered the scene with a groundbreaking claim: language models trained on the internet have reached their limits. From now on real scientific discoveries can only be achieved through experiments conducted in the real world. Therefore the company’s goal is to build “artificial scientists”—AI systems that formulate hypotheses conduct experiments learn from results and iteratively repeat this cycle to push the boundaries of knowledge.

Until now AI development has largely relied on text available on the internet. Although the volume of data available online is vast it is still finite. As Fedus notes even the most advanced models have largely exhausted this source. Reading a textbook many times may yield new perspectives but eventually discovery without experimentation is impossible. Periodic Labs’ vision is precisely to fill this gap by enabling computers to learn not just from words but directly from the laws of nature.

The company’s strategy is to build autonomous laboratories. In these labs robots will automatically carry out thousands of materials science experiments. Each experiment will generate gigabytes of unique data recording not only successes but also failures often omitted from scientific publications. AI will then analyze this data to determine how to design the next experiment. Thus the scientific process will become a machine-accelerated feedback loop.

Periodic Labs has already secured over $300 million in investment and reached a $1 billion valuation. Moreover it has recruited more than twenty researchers who left giants like OpenAI Meta and Google DeepMind. Thus this initiative is drawing attention from the scientific community through both financial power and access to top talent.

Yet despite its captivating vision this approach carries serious risks. First there is the issue of automating science. Human reasoning based on trial and error curiosity and intuition is being replaced by algorithms. This could transform the role of scientists from creators of new ideas to mere “overseers” monitoring machine-generated outputs.

The second risk is data monopolization. The massive data generated by these experiments will be collected by private companies like Periodic Labs. This could shift science from a public endeavor to a domain restricted by commercial interests. Historically scientific progress has depended on sharing experimental results and allowing others to verify them. If results remain confined to corporate databases this could deepen the unequal distribution of scientific knowledge.

The third issue is strategic dependency. Even a small advancement in superconductors or chip efficiency could provide major advantages in global competition. Therefore governments and major tech companies are likely to view such autonomous laboratories as strategic weapons. AI-assisted laboratories may become the new frontier of economic and geopolitical competition.

Periodic Labs is opening a new door at the intersection of AI and science. If this vision succeeds it could help humanity overcome limits in energy materials and technology. But it will also trigger serious debates regarding the autonomy accessibility and ethics of science. When science is entrusted to machines to accelerate progress will the resulting knowledge truly benefit all of humanity or will it become a new instrument of power in the hands of a narrow group?

The path opened by Periodic Labs may represent one of AI’s most ambitious dimensions: not merely generating content but establishing a new relationship with nature itself through AI. How we manage this relationship will determine the future of science in the coming years.