cld9032

It can help realize the potential of GAI more fully among different groups.

They are the motive force of further change and development.

Contradictions exist between subjects (users) and the tool (GAI), as well as between rules and the subject, and between rules and the community.

GAI users, including individual and organizational users.

GAI technology holds tremendous potential for application in various contexts and is a milestone for artificial general intelligence.

To analyze challenges and research directions associated with Generative AI from a socio-technical perspective.

Will GAI widen or bridge the digital divide among countries in different stages of technological development?

The development and advancement of GAI make current rules inadequate or inappropriate.

Accumulated structural tensions within and between activity systems heightened by ongoing transitions and transformations.

GAI can fabricate information and generate misinformation, referred to as hallucinations.

GAI should be viewed as an integral component of a larger socio-technical system.

The difficulty in distinguishing between GAI-generated content and human-generated content.

Users need specific skills and capabilities to effectively engage with GAI technologies.

A foundational understanding of AI concepts and ethical considerations is essential.

Users may resist GAI technology if they feel it takes over too much control of their lives.

They are linked systems that can influence or conflict with the central activity system of GAI usage.

The rights and responsibilities of users and the community.

Low download Internet speed, lack of Internet availability, and absence of complementary technology infrastructures.

They gain competitive advantages over companies that do not.

Factors influencing user acceptance include organizational context and individual user characteristics.

Unintended consequences that need to be examined.

Object, rules, community, division of labor, and outcome.

GAI can be utilized to tackle issues like environmental protection and Sustainable Development Goals.

Violent or offensive content, despite rules to avoid such outputs.

Human-AI relationship.

The ability to generate new content, high generalization capability, and reinforcement learning based on human feedback.

Human ethics, regulations, and laws related to the use of GAI.

Contradictions within each element, between elements, between activity systems, and between the activity system and its environment.

Limitations in transparency and explainability may fail to meet users’ expectations.

Applying the activity system framework to analyze activities and contradictions in a complex system.

Skills and capabilities that need to be identified and developed.

Policies should ensure fairness, transparency, and accountability in GAI applications.

Access to GAI technology may vary across countries or regions, leading to unequal usage.

Technological, ethical, societal, economic, regulatory, and governance challenges.

It may make users feel a loss of control and create technological resistance.

To learn a universal representation that can transfer with little adaptation to a wide range of tasks.

They are used to train a reward model based on reinforcement learning.

It may lead to the replacement and displacement of certain jobs and the emergence of new jobs.

The lack of transparency of GAI models makes it difficult to assess their output for accuracy.

It can converse, reason, remember context, and generate content such as poems, songs, and codes.

Unethical use and propagation of misinformation.

They may use self-attention in transformers and leverage unlabelled data for universal representation.

It allows the model to adapt to previously unseen data and perform well across a wide range of tasks.

Regulations should be adapted to address the unique challenges posed by GAI technologies.

GAI may widen the income gap between those who can upgrade their skills and those who cannot.

GAI requires manually labeled data, which restricts its applicability in domains with limited annotated data.

How can GAI play a role in improving human-human collaborations?

Studying ambulance service systems in hospitals.

GAI can produce content that appears real but is entirely fabricated, leading to misinformation.

To generate content by pre-training on a large corpus of data and then fine-tuning for specific tasks.

By better training students to utilize GAI effectively.

It has been studied and applied in areas including human-computer interaction.

Successful implementation requires clear strategies, user training, and organizational support.

Art creation, commercial advertisement generation, and long context-specific conversations in chatbots.

GAI can enhance personalization and efficiency in consumer interactions.

Contradictions within the elements, between the elements, between more advanced activity systems and the central system, and between different activity systems.

Existing laws and regulations on GAI are still inadequate and may disrupt current usage.

Conflicts between new and old activity systems as GAI tools become more advanced.

Current laws and regulations are inadequate to account for new phenomena brought about by GAI.

Analyzing human activity situated in a collective context.

Prompting strategies and fact-checking.

Improving transparency and explainability.

Biases can occur when training data only represents a fraction of the population, leading to exclusionary norms.

GAI could disrupt industries like tourism, e-commerce, arts, and entertainment.

Content generation, generalization ability, and reinforcement learning based on human feedback.

Impact on the future of work, effective collaboration with humans, and improving transparency while mitigating biases and misinformation.

It provides a meaningful framework to analyze trends and research directions GAI may bring.

Users’ motives to utilize GAI, such as increasing work efficiency or getting inspiration.

What are the obligations of GAI providers?

Determining appropriate authorship or acknowledgment associated with content jointly generated by GAI.

Do people believe in AI-generated news and human-written news to the same degree?

GAI models can generate original content by learning patterns from training data without explicit guidance.

Training programs should focus on ethical usage and responsible practices.

The need for algorithm transparency versus the pursuit of higher performance.

Biases may arise from the training data and processes used in developing GAI.

GAI may replace jobs, particularly low-skilled positions, leading to increased unemployment.

Users and GAI.

Object activities, tool-producing activities, subject-producing activities, and rule-producing activities.

It helps the model better align its output with users' needs and intentions.

The social groups that the user belongs to or other social actors that influence GAI usage.

Original content generation.

Human activity is object-oriented and achieved through actions using tools.

By calculating the probabilities of subsequent word-level tokens based on previously generated tokens and context.

Finding the best interface method to improve user adoption.

The gap in Internet skills and usage across different regions.

Artifacts/products that incorporate GAI technology.

GAI models have limited explainability and transparency, making it hard to interpret their outputs.

The execution results of the activity, which may include unexpected results like work efficiency and AI literacy.

Whether it promotes ethical use of GAI or leads to unethical use.

How can we detect AI-generated content with accuracy?

Data gathered from web pages may include personal information, threatening privacy.

Individual or organizational users deploying GAI as a tool to achieve certain objectives.

Balancing data collection needs with user privacy and data security.

Lev Vygotsky.

Content generation, generalization ability, and reinforcement learning using human feedback.

The optimization may conflict with ethical principles, leading to biases in the models' output.

Artificial data that captures the structure of real data sets rather than being extracted directly from the real world.

GAI may change workforce compositions rather than lead to outright job losses.

By creating realistic but false content, such as DeepFake images and videos.

It attracted worldwide attention as a Large Language Model that integrates the Generative Pre-trained Transformer (GPT) architecture.

Whether GAI will widen or bridge the digital divide.

This is a key research question to explore.

GAI may replace certain jobs while creating new opportunities in various sectors.

Accuracy, originality, and impartiality.

Human-ML augmentation, where humans and algorithms collaborate in decision-making.

GAI may streamline information search and influence decision-making behaviors.

Activity theory.

Shifting the focus from one-user interaction to the social context and real-world applications.