A New Collaboration with Technology
The evolution of technology is fundamentally transforming how we work. What was once merely a tool, AI is now evolving into a trusted partner. Like the relationship between a pilot and co-pilot in an aircraft, a new style of human-AI collaboration is emerging.
The Evolution of AI: From Tool to Partner
Traditional AI: A Commanded Entity
Previous generations of AI operated solely within the boundaries of detailed human instructions. For example, teaching an image recognition AI “this is a photo of a cat” and having it classify new images accordingly. It functioned much like an assistant following a strict manual.
New Era AI: An Autonomous Partner
A major transformation has been occurring from 2024 to 2025 with the emergence of “agentic AI.” Indeed, Claude 3.5 Sonnet was first announced in June 2024, followed by the upgraded version (Claude 3.5 Sonnet New) with computer use capabilities in October 2024. Google announced Gemini 2.0 Flash Experimental in December 2024, and OpenAI released Operator on January 23, 2025. The practical implementation of AI agents has been rapidly advancing during this period.
These AI systems can autonomously develop plans for given objectives, select necessary tools, and execute tasks. Specifically, they repeat four steps: “task planning, information gathering, reasoning, and action execution” to make autonomous decisions and work toward achieving goals.
For example, in response to a request like “Please prepare materials for next month’s sales meeting,” AI automatically carries out the following process:
First, it analyzes past meeting materials and collects sales data from internal databases. Next, it investigates competitor trends and creates graphs and charts from the obtained information. Then, it integrates these elements into a presentation and sends review requests to relevant stakeholders.
What’s important here is that humans only perform the initial goal setting and final confirmation, while intermediate processes are “delegated” to the AI. However, it’s worth noting that at present, complete automation of all processes has some limitations. AI agents may pause when encountering complex interfaces, password fields, or CAPTCHA verification, requiring human intervention.
A New Collaborative Style: Captain and Co-Pilot
Imagine an aircraft cockpit. The captain (human) is responsible for overall direction and final decisions, while the co-pilot (AI) supports complex calculations, real-time data analysis, and efficient operations. Just like in actual aircraft, humans bear the final decision-making authority and responsibility, while AI supports those decisions.
This relationship forms the core of international AI governance as the principle of “Human-in-the-Loop (HITL).” International standards such as ISO/IEC 42001 (AI Management Systems) require clear human oversight and accountability in critical decision-making processes.
New Skills Required of Humans
AI Management Capabilities
The ability to set appropriate goals for AI agents and evaluate their results becomes crucial. According to experts, this capability is considered one of the most important skills for the next decade. While similar to managing subordinates, it requires a deep understanding of AI characteristics. This includes understanding AI’s capabilities and limitations, appropriately assigning tasks, and assessing the quality of outputs.
Essential Problem-Discovery Ability
While AI excels at finding solutions, identifying “what should be solved” remains a human role. Deeper insights and creative problem formulation are required. Particularly important are the ability to translate business challenges into technical requirements, judgment to evaluate ethical impacts, and decision-making capabilities with understanding of organizational context.
Practical Implementation Approaches
A phased approach is effective when introducing new AI collaboration systems.
Starting with Small-Scale Pilot Programs
Rather than company-wide implementation, begin with areas where clear results can be easily measured, such as routine report creation or data analysis. In Panasonic Connect’s case, they deployed ConnectAI to all 12,500 employees in February 2023. In the first month after implementation, 55,380 questions were submitted (approximately 2,000 per day), and usage has since expanded, with about 5,000 questions now being asked daily.
An interesting aspect of this case is that while IT and technical staff were initially expected to be the main users, in reality, adoption was widespread across departments including legal and accounting. Additionally, adoption was seen across generations, with use ranging from new employees to senior staff.
Horizontal Deployment of Success Patterns
Deploy successful models from confirmed effective areas to other departments. The key is not just tool implementation but a review of entire business processes. For example, at Panasonic Connect, a legal department staff member who previously spent an hour reading long legal documents now spends less than 10 minutes reading AI-generated summaries. In the IT department, analysis of free-text responses to employee IT surveys, which previously took a team of three more than a week, now takes one hour using ConnectAI.
Continuous Improvement Cycles
AI agents learn and improve their performance with use. Regular evaluation and feedback are important. By continuously tracking usage, collecting user feedback, and upgrading systems as needed, the effectiveness of AI collaboration can be maximized.
Responding to AI Regulations and International Standards
International Regulatory Trends
With the rapid development of AI technology, AI regulations are being developed worldwide. In August 2024, the EU AI Act came into force and is being implemented in stages from 2025 to 2026. This law classifies AI systems according to risk levels and imposes strict requirements on high-risk systems.
In the United States, the Federal Trade Commission (FTC) and the National Institute of Standards and Technology (NIST) are providing AI risk management frameworks. In Japan, the Digital India Bill was announced in 2024, and the National AI Strategy Framework is expected to be developed in 2025.
Importance of International Standards
On World Standards Day in October 2024, ISO (International Organization for Standardization), IEC (International Electrotechnical Commission), and ITU (International Telecommunication Union) announced the 2025 International AI Standards Summit. The summit is scheduled to be held in Seoul, South Korea, on December 2-3, 2025.
These movements directly respond to the UN’s “Governing AI for Humanity” report’s call to action. The G7 Hiroshima AI Process has defined best practices for generative and frontier AI systems, and the Council of Europe’s 2024 Framework Convention on Artificial Intelligence became the first legally binding treaty focusing on human rights, democracy, and the rule of law in AI use.
Key International Standards
| Standard Number | Name | Content |
|---|---|---|
| ISO/IEC 42001 | AI Management Systems | Specifies requirements for establishing, implementing, maintaining, and continually improving responsible AI systems in organizations |
| ISO/IEC 23053 | Framework for AI Systems Using Machine Learning | Provides conceptual framework and common terminology for machine learning-based AI systems |
| ISO/IEC 5259 Series | Data Quality | Comprehensive standards for ensuring high-quality data for analytics and machine learning |
| NIST AI RMF | AI Risk Management Framework | Promotes trustworthiness, safety, and fairness in AI system development and deployment |
Actions Organizations Should Take
Organizations implementing AI need to be mindful of the following points:
Establishing risk assessment and governance structures is essential. By building AI management systems compliant with international standards such as ISO/IEC 42001, organizations can expect global compliance, smooth cross-border operations, and enhanced market trust.
Ensuring transparency and explainability is also important. Organizations need to document AI system decision-making processes and establish systems to explain them to stakeholders. This includes creating technical documentation, managing risk logs, and post-market monitoring.
Clarification of human oversight and accountability is required. It’s necessary to clarify the ultimate human responsibility in critical decision-making and establish appropriate oversight systems. Particularly for high-risk AI applications, human approval processes may be mandated.
Attention must be paid to data governance and privacy protection. Ensuring the source, quality, and ethical acquisition of data used by AI systems, and complying with regulations on personal data protection (such as GDPR) is important.
Implementing continuous education and training is also indispensable. Providing education to employees on appropriate use of AI technology, potential risks, and ethical considerations, and improving AI literacy throughout the organization is necessary.
Conclusion: A Partnership of Mutual Growth
Collaboration with AI holds meaning beyond mere technological innovation. It represents a new form of value creation where human creativity merges with AI’s analytical capabilities.
What’s important is directing the time and energy created by “delegating” to AI toward more creative and human activities. Leveraging technology to pursue essential value that only humans can provide—that will be the key to thriving in the coming era.
However, this new relationship also carries responsibility. Organizations are required to comply with international standards and regulatory requirements, maintain ethical considerations, and utilize AI while protecting human dignity and rights. By adopting international standards such as ISO/IEC 42001 and establishing transparent governance structures, organizations can build a foundation for trustworthy AI collaboration.
AI is the co-pilot, we are the captains. Together, we will fly higher and farther toward the future. However, this flight must always be guided by human values and ethical judgment, conducted according to international rules and standards.
Comment