Is “AI Dependence” a Bad Thing? Data Reveals a New Reality
The figures from ASCII.jp’s survey are striking. Approximately 70% of company employees acknowledge that they “cannot do their jobs without generative AI.” Simultaneously, 90% of CEOs are confident that “AI agents will produce results within the year.” This gap vividly illustrates the current state of corporate AI adoption, where on-the-ground “dependence” and management’s “expectations” are colliding.
However, the question we should be asking is not “Should we reduce dependence?” but rather “How can we transform this structure of dependence into a management asset?” This media platform advocates for a perspective that views AI not merely as a convenient tool, but as an “extension of human resources” that fundamentally alters the reproducibility and scalability of business operations.
The Concrete Strategy Demonstrated by Credit Saison’s “Company-Wide AI Worker” Initiative
The case of Credit Saison, reported by Nikkei Business, can be seen as a pioneering example of embodying this perspective. The company has announced a plan to transform all approximately 3,700 employees into “AI Workers,” anticipating efficiency gains equivalent to 1,500 personnel. This is on a completely different dimension from simply introducing ChatGPT internally.
What deserves attention is their concrete approach. According to reports, the company established an “AI Utilization Promotion Office” to build a cross-departmental framework. While conducting training for all employees, they are also adapting business processes themselves for AI integration. This is a strategic investment aimed not at sporadic tool adoption, but at raising the organization’s overall “AI readiness capability.”
The “Hidden Champions” Holding Another Key to Generative AI Supremacy
Meanwhile, the “Japanese hidden companies that even GAFAM respects,” as pointed out by Business+IT, refer to semiconductor manufacturing equipment and materials makers that support the foundation of generative AI. While OpenAI and Google fiercely compete in model development, these companies supply the essential hardware and materials, building an unshakable position.
This structure offers a crucial insight for corporate AI strategy. In other words, becoming a “frontrunner” that develops large language models in-house is not the only path to success. For many Japanese companies, a more realistic strategy is to leverage existing strengths and resources to identify an “indispensable role” in the AI era.
Three Concrete Actions Management Should Take
Based on these news items, the actions that executives and CTOs should consider immediately can be summarized in the following three points.
1. Visualizing “AI Dependence” and Strategically Enhancing It
It’s time to move beyond fearing employee AI dependence and start measuring and managing it. Based on our client case studies, we evaluate using the following framework.
- Level 1 (Assistance): One-off tasks like proofreading, information gathering.
- Level 2 (Collaboration): Routine tasks like data analysis, automated report generation.
- Level 3 (Integration): Core tasks like decision support, automated customer response.
First, understand your company’s “average AI dependence level.” Then, prepare the training and tool environment to elevate that level by one stage. Starting with basic training for all employees, as Credit Saison did, is the first step to prevent knowledge silos and generate organizational strength.
2. Investing to Amplify Your Company’s “Hidden Strengths” with AI
Not every company needs to enter the AI model development race. Instead, priority should be given to investments that use AI as an amplifier for the “domain knowledge,” “customer touchpoints,” and “business processes” your company has accumulated over many years.
As a concrete example, here is a case of a mid-sized manufacturer we supported. The company digitized its product adjustment and maintenance manuals (a treasure trove of non-digital data) by linking its in-house expert technicians with ChatGPT Enterprise. They built a Q&A-style knowledge base. The initial investment was approximately ¥3 million (including AI consulting and environment setup), but it yielded results including a 40% reduction in customer response time and a 60% reduction in new employee training periods.
The core of this investment was not the introduction of expensive SaaS, but the construction of an internal process to “translate” the company’s tacit knowledge into a format AI can process.
3. Environment Preparation and Cost Design for “AI Workers”
For all employees to utilize AI, environmental preparation beyond mere tool provision is necessary. The Credit Saison case suggests they are comprehensively preparing the following elements.
- Integrated Access Environment: Single sign-on for multiple AI tools (ChatGPT, Claude, Copilot, etc.)
- Security Guidelines: Handling of confidential information, verification process for generated results.
- Internal Best Practice Sharing Platform: Templatizing and internally deploying successful use cases.
Regarding costs, the following models based on our experience can serve as a reference.
- Small-scale Introduction (~50 users): Approximately ¥100,000~ per month for ChatGPT Team plan ($25/user/month) + Claude Team ($30/user/month). Initial consulting fees separate.
- Mid-scale Introduction (~500 users): Considering volume discounts via Enterprise contracts and potential use of proprietary APIs. Anticipated monthly costs of roughly ¥1 million to ¥5 million.
- Large-scale / Industry-Specific: Development of fine-tuned models using proprietary data or custom solutions. Initial investment from ¥10 million~, but capable of building long-term competitive advantage.
The crucial point is to design a management cycle that not only considers tool costs but also reinvests the human resources freed up by AI into “higher value-added tasks.”
From AI Dependence to AI Co-Creation: The 2024 Watershed
The situation where 70% of employees feel dependent on AI signifies a point of no return. The issue is not dependence itself, but whether that dependence remains at the level of “passive use” or evolves into “active co-creation.”
Credit Saison’s “AI Worker” transformation is a pioneering case attempting an organizational shift towards the latter. Furthermore, the Japanese companies demonstrating presence in the hardware foundation of generative AI teach us the importance of redefining one’s core competencies within the context of the AI era.
The next move for management is to turn both wheels simultaneously: preparing the environment to “enable” AI use and fostering a culture where employees “co-create” with AI. Beyond that lies the “results produced by AI agents” that 90% of CEOs in the survey believe in—namely, a dramatic leap in productivity, a focus on creative tasks only humans can do, and a new reproducibility in management.
Dependence is not the end. It is merely the beginning of a new collaborative relationship between humans and AI.
Comments