Usage saturation and context dilemma, what the trial of four hundred Saudi media professionals with generative tools reveals

Professional frequency hypotheses collapse before a single figure in the Saudi audio-visual production sector: 95.8 percent of practitioners use AI tools in content creation. A survey of 400 specialists published in Discover Artificial Intelligence reveals that only 17 professionals remain outside this wave, shifting the question in newsrooms and editing studios from technology adoption to the depth of its use, its safety level, and its regulatory fairness.

The research conducted in 2025 was led by an academic team headed by Saad Faraj Al-Anzi and Ahmed Tawalba of Gulf University in Bahrain, with the participation of Farhat Yilmaz and researchers from Saint Petersburg University. The team built its analytical framework by integrating three theoretical strands: the technology acceptance model, the digital divide theory, and institutional theory. This approach concludes that barriers and enablers are two sides of the same coin; cost becomes an obstacle when subscriptions are expensive, and a catalyst when software is affordable for production teams.

Privacy is not solved by purchasing subscriptionsAmong the limited group that has not adopted the tools, security and data-protection concerns topped the list of obstacles, with 12 of the 17 professionals citing them, followed by high subscription costs for 11 respondents. The researchers explain this disparity by noting that visual and audio media workers handle highly sensitive material, including raw footage not yet broadcast, pre-publication marketing assets, client data, and source identities. Cloud-based AI tools usually require these materials to leave institutional control for processing, at which point the Saudi personal data protection system intervenes, making the consequences of any mis-estimation concrete and immediate, beyond theoretical concerns.

In contrast to Western literature that highlights journalists’ anxiety about job loss, fear of occupational substitution appeared at the bottom of the list and was mentioned by only two participants among the non-users. The researchers attribute this to the expansion of the media sector within Vision 2030 goals, where participants view AI as a means to boost efficiency and foster innovation rather than as a substitute threatening their job security. The full sample also showed that specialized training programmes and workshops were the most sought-after enablers, followed by the provision of user-friendly tools at reasonable prices, and high-quality training data and open-source resources.

Demographic distributions reveal striking contradictions within the local market, as AI readiness varied by age, occupational sector, region, and income, while no statistically significant differences were recorded by gender or education level. The government media sector topped the readiness scores, driven by digital-transformation directives and the accompanying provision of tools, training, and work hours allocated for learning and experimentation.

Field experience raises the ceiling of technical critiqueIn the geographic comparison, the central region, which includes Riyadh and the kingdom’s most extensive media infrastructure, recorded the lowest average readiness, while the southern region topped the indicators. The researchers offer a technical and social explanation for this disparity: central-region practitioners have the greatest cumulative exposure to the tools in real production settings, and therefore have directly encountered model failures, especially in understanding Arabic linguistic context, which ranked second among the main technical challenges after the lack of large datasets. Conversely, the readiness of southern practitioners reflects the intense demand to produce promotional and tourism content that aligns with major regional projects, where automation provides a quick solution to offset operational capacity gaps.

The linguistic dilemma constitutes a structural technical knot that exceeds the capacity of training courses to address individually; Arabic carries morphological richness and a range of registers from Modern Standard Arabic to local dialects, as well as the absence of diacritics in written texts. Models trained on foreign-language data struggle with these specificities, leading to errors in dialect tone detection, accurate transcription, and contextual disambiguation. Accordingly, the researchers recommend that national AI agencies fund the development of language-specific models and resources for the media industry, and provide licensed, reliable Arabic audio and video corpora.

For CEOs or media decision-makers in the Gulf, Egypt, and the Levant, the editorial board sees these findings reshuffling spending priorities. The race is no longer about introducing generative tools, as most production rooms have already adopted them, but about moving from superficial individual use to building production lines that comply with data-governance requirements and local regulations. Note that the next operational challenge lies not in mastering prompt writing, but in crafting regulatory controls that prevent sensitive material from leaking outside the work environment, and in activating strict human review that catches model shortcomings in contextual and dialect understanding before content is broadcast to the audience.