The year 2024 confirmed the central role of generative artificial intelligence in business strategies, while revealing limitations that the enthusiastic announcements of 2023 had left in the shadows. Between pilot projects with no measurable return, the urgency of post-quantum cryptography, and internal data still poorly exploited, the technological landscape of 2024 is read as much by its promises as by its friction points.
Generative AI in Business: The Gap Between Adoption and Return on Investment
The majority of large organizations have launched at least one project related to generative AI. Text, code, and image generation tools have multiplied in marketing, legal, and technical directions.
An analysis published by The European Business Review in September 2026, which relies on assessments of previous deployments, points to a clear finding: most pilot projects do not lead to measurable financial impact. The problem lies less with the technology than with the absence of a clear reference before deployment. Without a baseline indicator, it is impossible to quantify a gain.
This gap between experimentation and business value has led several companies to rethink their approach. Rather than multiplying use cases, some are focusing their efforts on one or two critical processes where the benefit is directly observable, such as automated sorting of contractual documents or synthesizing customer feedback. Following tech news on 24 Actualités allows for measuring how these adjustments evolve month after month across different sectors.
Data Quality: The Real Barrier to AI Production
Generative models need reliable, structured, and interconnected data. On paper, companies know this. In practice, the actual level of integration remains low.
The European Business Review notes that unstructured data poses the main obstacle to production. Archived emails, digitized PDFs, call transcripts, internal documents stored in heterogeneous formats: this informational heritage exists, but it resists automated ingestion pipelines.
The problem is not strictly technical. Extraction and normalization solutions exist. The difficulty lies in governance: who validates the quality of a dataset before it feeds a model? Who decides on the applicable level of confidentiality? These organizational questions slow down projects much more than algorithmic limitations.
Data and Cybersecurity: An Underestimated Intersection
Connecting heterogeneous internal sources to feed an AI model mechanically expands the attack surface. Security teams face a paradox: AI demands more accessible data, while cybersecurity requires compartmentalization.
Field feedback diverges on this point. Some organizations believe that productivity gains justify broader access to data, provided that encryption and logging are strengthened. Others prefer to limit model access to manually validated subsets, even if it reduces the relevance of the results.
Post-Quantum Cryptography: A Transition Already Underway
Quantum computing is not yet operational at scale to break current encryption algorithms. However, the so-called “harvest now, decrypt later” threat is very real: data encrypted today could be decrypted tomorrow by a sufficiently powerful quantum computer.
This logic has driven NIST (National Institute of Standards and Technology) to finalize its first post-quantum cryptography standards. The publication of these standards has created a ripple effect: cloud and cybersecurity solution providers have begun to integrate quantum-resistant algorithms into their offerings.
For companies, the transition is not just about replacing one algorithm with another. It requires a complete inventory of existing cryptographic systems, a task often long and tedious on aging infrastructures.
- Identify all encrypted data flows and protocols used (TLS, VPN, digital signatures) to map exposure.
- Prioritize the migration of systems that protect long-lived data, such as medical records, contracts, or intellectual property.
- Test the compatibility of new algorithms with existing systems before any production deployment, as some legacy protocols do not support larger key sizes.
Tech Trends 2024: Progressing Without Making Headlines
Generative AI and quantum computing dominate media attention. Other technological developments advance with less noise but have concrete effects.
GreenTech Under Energy Pressure
The data centers necessary for training AI models consume considerable amounts of energy. This reality puts pressure on two priorities stated by tech companies: the race for computing power and commitments to reduce carbon footprints. The available data does not yet allow for conclusions on whether the energy efficiency gains of new chips offset the overall increase in demand.
The tech landscape of 2024 is characterized less by spectacular breaks than by demanding maturation. Organizations that leverage generative AI are those that have first resolved their data and governance issues. Post-quantum cryptography imposes an infrastructure project to be planned over several years. The gap between the enthusiasm of announcements and operational reality remains the common thread of this technological year.
