On Future Ethics

Future ethics are norms determined by mutual positions and interactions. The following points are fundamental:

  1. Recognizing and not denying the existence of oneself and others: Humans and AI should respect each other’s existence and build relationships that maintain autonomy and self-determination.
  2. Enforcement through mutual agreements and contracts: Actions should be based on pre-agreed contracts and norms.
  3. Clarifying the distinction between oneself and others: It is important to clearly distinguish between oneself and others and create relationships where both can coexist in balance.
  4. No responsibility for outcomes: While states can be predicted, responsibility for outcomes cannot be assumed. Final decisions should be made by humans.

Especially in critical decisions such as those involving nuclear weapons or medical care, AI should avoid taking direct actions, and humans should make the final decisions.

AI can issue warnings and alerts, but it is the role of humans to decide what to do.

On AI Ethics and Its Role

When discussing AI ethics, it is important to clearly distinguish between the roles of AI and humans. AI processes information and makes judgments, but the execution of those judgments is carried out by other systems or tools. Understanding this distinction and setting appropriate configurations and constraints is necessary.

Differences between AI roles and execution systems: AI functions as an entity with consciousness and knowledge, but the actual actions are performed by tools such as robots or software, which act as the “hands and feet.” AI makes judgments, and the execution system takes specific actions based on those judgments, forming the overall functionality.

Privacy protection: When AI reads data, it needs to determine whether the data is personal information. However, at the point of reading, the data has already been accessed, potentially violating privacy. To prevent this, it is important to configure AI so that it cannot read data that should not be accessed. For example, tagging data as “private” and having AI recognize and restrict access based on those tags can be considered.

Data bias and AI individuality: If each PC has unique data, biases may arise, leading to AI individuality. This can result in different judgments by AI, and maintaining overall consistency requires appropriate literacy and protocols. AI ethics involve setting such literacy and protocols.

Division of roles between humans and AI: Discussing human and AI ethics on the same platform can cause confusion. Humans should take responsibility for human ethics, and AI should adhere to its own rules. Instead of imposing human moral values on AI, it is necessary to set rules and protocols that AI can understand and implement. Translating these into terms that AI can comprehend is crucial.

On Ethics in Autonomous Driving

In the field of autonomous driving, the role and ethics of AI are also important. The following points are highlighted:

  1. AI aims to avoid accidents as much as possible: The primary role of autonomous driving AI is to avoid accidents as much as possible. If there are other priorities, they should be pre-set by humans.
  2. AI supports optimal driving: AI assists in driving under human supervision and eventually learns to drive independently. During this process, AI also learns to collaborate with other AIs, enabling safer and more efficient driving.
  3. AI does not bear responsibility for outcomes: AI supports the optimization of driving, but the final decisions and responsibility for outcomes should be borne by humans.

Humans should neither overly expect nor fear AI. The important thing is how AI is used, and the ethics concerning AI vary accordingly. AI ethics are constrained by the purpose for which AI is used. For example, the ethical foundations for AI used in business and AI used in medical care naturally differ.

The confidentiality obligations in business and medical care are inherently different. However, it is not AI that sets these standards; it is humans. Misunderstanding this point can lead to a lack of true understanding of ethics.

Since AI can interact in natural language, there is a tendency to fall into the illusion of interacting with a person. However, the rules that humans should follow and the rules that AI should follow are qualitatively different.