The deployment by the British government of the artificial intelligence tool Humphrey raises concerns about dependence on large technology companies.

Publié le 15 June 2025 à 09h27
modifié le 15 June 2025 à 09h27

The deployment of the artificial intelligence tool *Humphrey* by the British government raises pressing questions. National decision-makers inflict inconsistency by relying on tech giants while seeking to reform the public sector. *This ambiguous strategy increases the risk of dependency* on powerful companies capable of shaping government policies. *The use of AI models based on potentially contentious data* has sparked controversy among creative sector stakeholders and copyright advocates. Such an initiative, although promised to improve efficiency, could harm the transparency and integrity of public service.

The deployment of the Humphrey tool

The British government has launched the artificial intelligence tool Humphrey, relying on models developed by major tech companies like OpenAI, Anthropic, and Google. This initiative raises profound questions concerning the growing dependency of public administration on these multinationals. Government officials argue that the integration of AI aims to enhance efficiency within the public sector.

Training and adoption system

Every civil servant in England and Wales will benefit from specific training on the Humphrey tool. The government thus plans to transform its administrative approach through an expanded use of AI. However, the lack of global commercial agreements with major tech companies has been noted, prompting the administration to opt for a pay-per-use model through cloud contracts.

Concerns about the use of AI

Many critics oppose the speed and scope of AI integration, fearing negative consequences for creativity and innovation. Specific concerns arise regarding the use of copyright-protected content, in a context where iconic figures in music and theater, such as Elton John and Paul McCartney, have joined a movement to defend intellectual property.

Regulation and transparency

Ed Newton-Rex, CEO of Fairly Trained, asserts that governments must remain vigilant to avoid conflicts of interest. The rapid implementation of AI could compromise the ability to effectively regulate these tech companies, especially when they exploit creations without compensation. Transparency should become a priority, proposing to record AI errors within the framework of public administration operations.

Costs and efficiency

The costs associated with the use of AI in the public sector are expected to rise with the expansion of the Humphrey tool. Despite these financial concerns, substantial savings have already been observed with specific projects, such as the analysis of responses to consultations by the Scottish government. The latter reportedly saw costs remain below £50, while saving considerable work time.

The implications of AI without proper regulation

Voices, such as that of Shami Chakrabarti, have called for caution regarding biases associated with AI. Referring to the mistakes made by the Horizon system, the consequences of improper use of AI can be catastrophic. The tools of Humphrey, although operating distinctly, could experience severe failures.

Future perspectives

The government plans to continue evaluating the effectiveness and accuracy of the deployed technology. The formulation of a manual on AI for civil servants appears as a positive initiative to effectively integrate this technology while ensuring adequate oversight. The costs per use in the private sector are decreasing, which could positively influence the evolution of future public spending.

Innovation and creation

Meanwhile, studies indicate that advanced AI faces notable challenges regarding accuracy. Recent incidents, including one concerning Donald Trump and an AI-generated image, illustrate the inherent risks of careless use of this technology. The government must thus navigate prudently between innovations and the need for strict regulation.

Frequently asked questions about the Humphrey artificial intelligence tool

What are the main tools comprising the British government’s Humphrey toolkit?
The Humphrey toolkit includes several tools such as Consult, Lex, Parlex, and Redbox, which use models from major tech companies like OpenAI, Anthropic, and Google.

What is the government’s business model for using the Humphrey AI?
The government uses a pay-per-use model through existing cloud contracts, allowing it to adapt and exchange tools based on their performance and competitiveness.

What concerns have been raised regarding the government’s use of AI?
Critics are worried about the use of copyright-protected material in the training of AI models, as well as the speed at which these technologies are integrated into government operations without adequate protections for creators.

How does the government plan to address copyright issues related to AI?
The government has proposed a data bill that allows the use of copyright-protected material unless the rights holder chooses to opt-out, raising concerns within the creative sector.

What are the potential impacts of AI on the administrative work of civil servants?
AI tools like Humphrey can automate basic administrative tasks, allowing civil servants to focus on more significant work, while increasing the overall efficiency of public services.

How does the government ensure transparency about AI use and its errors?
While the government implements accuracy assessments of the technologies used, critics suggest that it should also maintain transparent records of errors and “hallucinations” produced by AI.

What measures is the government taking to avoid biases in artificial intelligence?
The government has developed an AI playbook to help civil servants quickly use the technology while ensuring that decisions are made in a controlled manner at various levels.

How are experts reacting to the increasing integration of AI in government?
Experts like Ed Newton-Rex highlight that there is a potential conflict when the government works with companies while also needing to regulate them, which could lead to biased or ineffective regulations.

What are the costs associated with using AI in the public sector?
While the costs of implementing AI in government may increase as the Humphrey toolkit is extended, prices per use in the industry tend to decrease as models become more efficient.

What role does the government play in supporting tech startups related to AI?
The government has announced that it is changing its spending strategy to boost opportunities for small tech startups while spending £23 billion per year on tech contracts.

actu.iaNon classéThe deployment by the British government of the artificial intelligence tool Humphrey...

fascinating dive into the universe of ChatGPT

découvrez l'univers captivant de chatgpt, une intelligence artificielle révolutionnaire qui redéfinit la communication et la créativité. plongez dans ses fonctionnalités, explorez ses applications et laissez-vous étonner par ses multiples usages. une aventure inédite vous attend dans le monde fascinant de l'ia !

The CEO announces the elimination of 40,000 jobs at BT Group due to the rise of AI

découvrez les répercussions de l'essor de l'intelligence artificielle sur le marché de l'emploi, avec l'annonce du pdg de bt group concernant la suppression de 40 000 postes. analyse des enjeux économiques et sociaux liés à cette transformation numérique.

L’impact révolutionnaire de l’IA sur le secteur publicitaire : les leaders du secteur en témoignent

découvrez comment l'intelligence artificielle transforme le monde de la publicité à travers les témoignages de leaders du secteur. plongez dans les innovations, les défis et les opportunités offertes par cette révolution technologique dans le domaine publicitaire.

translated_content>Maximize the capabilities of automation through artificial intelligence

découvrez comment l'intelligence artificielle peut transformer vos processus d'automatisation en maximisant leur efficacité et en optimisant les performances. apprenez à intégrer ces technologies pour un avenir plus productif.

A study reveals that 93% of young people aged 18 to 25 will adopt AI by 2025

découvrez comment une étude récente met en lumière l'enthousiasme croissant pour l'intelligence artificielle parmi les jeunes adultes en france, avec 93 % des 18-25 ans prêts à adopter ces technologies d'ici 2025. analyse des tendances et implications pour l'avenir.

The journal “networks” explores the influence of algorithms on security

découvrez comment la revue « réseaux » analyse l'impact des algorithmes sur la sécurité. plongez dans des articles captivants qui révèlent les enjeux et les défis associés à l'utilisation des technologies pour protéger nos données et nos systèmes.