#46 Society and Technology

The rapid advancement of artificial intelligence technologies has shifted the focus of experts in mathematics and computer science toward large language models. However, Elon Musk’s AI venture xAI is breaking the mold with its latest job posting for Grok. Departing from convention, the company is now seeking award-winning writers, master journalists, and experienced screenwriters rather than engineers. This posting aims to elevate Grok’s literary quality and narrative power by encompassing high-quality content creation, editing, and text analysis processes. The most talked-about detail on social media is the high salaries offered to these writing masters.

Large language models generate new content by training on data compiled from existing text corpora, using novel combinations of the training data. In models built on the structural use of language, the quality and scope of the dataset serve as the primary variable determining the characteristics of the model’s output. For instance, a model trained exclusively on 19th-century literature will inevitably produce outputs reflecting the stylistic and thematic traits of that era. In this context, the optimization and training of language models are no longer confined to the technical boundaries of computer engineering; they have also become central research priorities in linguistics, social sciences, and creative writing disciplines.

xAI’s job posting seeking award-winning writers and journalists is not unprecedented in this regard. It is plausible to say that other language models have also collaborated with text writers for fine-tuning. However, several critical issues stand out. First, there is a significant risk that the essential qualities of writers’ labor and high-quality writing activities will be reduced to mere “data input.” Writing is simultaneously an activity with multiple purposes: it is not only a vehicle for conveying the writer’s ideas but also a vessel for preserving them, presenting them through specific forms and stylistic embellishments.

As a tool that unites the content and form of ideas, writing enhances the effectiveness of ideas when it emerges from skilled hands, leveraging stylistic and expressive possibilities. From this perspective, another danger for a text-generation model like Grok lies in its tendency to render radical or extreme views palatable by cloaking them in the refined language it has acquired through training. The embellishment of manipulation, transformed into language already familiar within the collective social memory, carries predictable risks. This is why the hiring criteria require candidates to be either award-winning writers or creators of content consumed by broad audiences.

From the writers’ perspective, potential scenarios also carry risks for the future of authorship and authentic writing. For a writer training a language model, there exists a paradox laden with professional tension: the very models they train may replace them, generate scripts that favor award-winning writers, or write stories themselves. In this context, writers risk being pushed out of the ecosystem with one-time payments.

All these tensions and risks will certainly not deter writers from contributing to language models. The future awaiting us may be one in which texts become homogenized, intellectual property and authorial text production are lost, and originality has long vanished. While it is essential to recognize this transformation, merely reacting against it is insufficient. In an era when our intellectual nourishment is turning to desert, we must not lose the oases we find, nor abandon them to the leveling force of the desert. Otherwise, we are very likely to be woven into texts constructed by statistical calculations that assemble letters and words—texts in which the author’s hand is absent, intent is missing, and what matters is not what is said but how it is said.

On January 9, the Pentagon released its artificial intelligence strategy for the U.S. Department of Defense. This brief six-page document underscores how critical AI will be for future military technologies and emphasizes the urgent need for the Department to adapt to these developments. Such strategy documents often appear to offer many insights but typically contain general statements and non-specific information. Of course, when evaluating these documents, we must acknowledge that classified portions remain undisclosed.

One significant point in the document is the existence of multiple threats against units within the Department that fail to adapt to AI. For example, under the subheading “AI-Based Warfare,” the document states: “We must abandon old approaches to warfare and use this disruptive technology to enhance the lethality of our armed forces. Exercises and experiments that do not meaningfully incorporate AI and autonomous capabilities will be reviewed by the Director of Cost and Program Evaluation for resource reallocation.” The term “review” here implies that budgets for such units may be cut. In other words, if AI usage in projects and programs is deemed insufficient, the Pentagon may reduce their funding.

Another notable point is found under the heading “Data Access,” which grants every employee within the Department the right to request data from the CDAO (Chief Digital and AI Office). The document mandates that all data held by employees—regardless of classification level—must be cataloged for the CDAO within 30 days and shared within seven days upon request. If data is not shared, the employee must first provide an explanation to the Under Secretary of Defense for Research and Engineering; if this is deemed inadequate, they must then explain directly to the Deputy Secretary of Defense. This effectively compels Department employees to share their data with the CDAO to avoid entering this burdensome bureaucratic process.

Another key focus is the removal of “barriers” encountered in project and operations management. These barriers primarily involve subjecting every operation on the Department’s network to cybersecurity protocols to ensure no malicious software is present. Although this protocol is necessary for security, it also slows the adoption of new systems and projects. The published document calls for a “wartime” attitude toward these barriers and suggests that risks should be accepted to accelerate processes. To this end, a council under the Office of Research and Engineering will meet monthly to eliminate obstacles, and various processes for data sharing without formal hiring will be expedited within the Department. This high-risk move reflects the Department’s recognition that AI adoption cannot be achieved through investment and incentives alone; internal processes must be accelerated as if in wartime.

When considered together, these three points indicate that although the document does not specify exactly how AI should be adopted or used, it clearly states that this adaptation must be enforced within the Department, disregarding certain obstacles, and that necessary steps must be taken immediately. This raises a pressing question: how will AI not only technologically but also directly influence future warfare?

After the majority stake in TikTok’s U.S. operations was transferred to American investors, the platform faced a serious legitimacy crisis not only over its ownership structure but also regarding content policies, data security, and allegations of political censorship. The new structure, established as TikTok USDS Joint Venture LLC, aimed to circumvent a potential U.S. ban, but trust in the platform rapidly declined following this move.

Many U.S. users and content creators began to believe that TikTok’s new ownership structure would lead to intensified surveillance and censorship. Updated terms of service and privacy policies more clearly defined the types of data collected, but this transparency generated unease rather than increased user trust. According to Sensor Tower data, in the days following the announcement of the deal, TikTok’s uninstall rates in the U.S. nearly doubled.

Beyond these technical and legal concerns, public backlash focused most intensely on the visibility of political content. Numerous users alleged that critical content regarding Israel’s massacres in Gaza and human rights violations in the West Bank was being suppressed. TikTok’s new CEO for U.S. operations, Adam Presser, defended these actions as part of combating hate speech, but in practice, this was interpreted as the systematic exclusion of criticism of Zionism. The suspension or restricted access of Palestinian journalists and activists’ accounts further reinforced this perception.

In this context, calls to leave TikTok evolved beyond individual responses into a collective digital boycott campaign. Many comedians, activists, and independent journalists did not merely delete their accounts but also urged their followers to migrate to alternative platforms. Even California Governor Gavin Newsom publicly commented on the suppression of anti-Trump content on the platform and signaled an investigation, demonstrating that the debate extended far beyond social media users.

The platform emerging as the leading alternative is UpScrolled. Founded by Palestinian-American entrepreneur Issam Hijazi, the platform positions itself as offering “transparent technology” and “censorship-free social media.” Following TikTok’s ownership change in the U.S., UpScrolled experienced a surge in downloads. The app climbed to the top of the Social Networking category on the U.S. App Store. Users viewed it as a genuine alternative to Big Tech, appreciating its chronological feed, algorithmic transparency, and commitment to user autonomy.

However, a critical handicap exists. Historically, digital platform shifts often begin as political responses but either persist through technical sustainability or collapse. The long-term success of UpScrolled and similar alternatives will depend not only on their ideological stance but also on their infrastructure capacity, moderation balance, and service continuity. Server crashes and scaling challenges remain among the greatest risks facing such alternatives.

Ultimately, the TikTok crisis goes beyond a single app controversy; it brings to the forefront the issue of control and freedom within the digital public sphere. User reactions are not merely directed at one company’s policies but at the centralized technological power structures operating on a global scale. Initiatives like UpScrolled are a product of this backlash; however, whether this dynamic endures depends on whether alternatives can genuinely deliver the freedom they promise in a sustainable manner.