Generative AI: An Assessment in Mid-2024

Introduction

The advent of Generative Artificial Intelligence (AI) has been hailed by many as a groundbreaking technological revolution with the potential to reshape industries, enhance creativity, and automate complex tasks. Proponents have envisioned a future where Generative AI dramatically boosts productivity and economic growth. However, a closer examination reveals a more nuanced reality.

The Promise of Generative AI

Generative AI, which includes technologies like deep learning models, natural language processing, and creative AI, has shown remarkable capabilities in specific applications. Tools such as OpenAI's GPT-4, Google Gemini and Meta’s Llama Models have demonstrated proficiency in generating human-like text, while AI-driven image and video generation have opened new potential avenues for digital entertainment. These advancements have led to high expectations about AI's potential to transform various sectors, from customer service to content creation.

Measured Productivity Gains

Despite the impressive capabilities of Generative AI, the anticipated widespread productivity gains have not fully materialized. Recent reports from the Bureau of Labor Statistics (BLS) indicate only modest improvements in productivity. For instance, in the first and second quarters of 2024, nonfarm business sector labor productivity increased by 0.3% and 2.3%, respectively (bls). These figures suggest incremental rather than transformative improvements.

A similar trend is observed in the OECD's productivity data. While specific sectors and larger firms have reported productivity gains due to AI adoption, the overall impact across all firms remains limited.

Sector-Specific Impacts and Firm Size Variations

The productivity benefits of Generative AI are not uniformly distributed. Larger firms with more resources to implement AI technologies have seen more significant gains. For example, a study in the UK found that while larger firms (with more than 250 employees) experienced productivity premiums from AI adoption, the effect was statistically insignificant when all firms were considered. This suggests that the transformative potential of AI is more pronounced in specific contexts rather than universally applicable.

Complementary Skills and Assets

One critical factor influencing the productivity gains from AI is the presence of complementary skills and assets. Firms that invest in high-skill professionals and advanced digital infrastructure can maximize the benefits of AI technologies. All reports, and our own experience with implementations of assistive AI technologies, show that the productivity benefits of AI are significantly enhanced when these complementary assets are in place.

Ethical and Societal Challenges

Beyond technical and economic considerations, ethical and societal challenges also constrain the transformative potential of Generative AI. Issues such as bias, privacy concerns, and the need for substantial computational resources pose significant hurdles. These challenges necessitate careful consideration and regulatory frameworks to ensure that AI technologies are deployed responsibly and ethically.

Case Studies and Real-World Applications

Real-world applications of Generative AI provide further insights into its actual impact. For instance third party studies show that a generative AI tool used in customer support increased productivity by 14%, primarily benefiting less experienced workers (Link). Similarly, AI tools that assist with writing tasks have shown productivity improvements, particularly for lower-skilled workers. Our own experiences with developing assistive AI tools show similar productivity gains with GenAI effectively helping in well understood areas of friction and even repetitive tasks that do not require significant creativity and problem solving.

Finally, it must be noted that AI – if not applied properly, especially in an assistive function – can quickly become hated and generate a net productivity loss and drive employee dissatisfaction. In the OECD study cited above one in five respondents reported that AI decreased their autonomy, and this share is significantly larger among the relatively small group who report to be subject to algorithmic management.

These case studies demonstrate that while Generative AI can drive productivity in specific applications, its broader impact remains limited and if its applications is not designed and implemented well, AI systems can increase work intensity and reduce productivity.

Conclusion

Generative AI still holds significant promise, but its role is currently more supplementary than revolutionary. The technology has contributed valuable tools and advancements, but the anticipated transformative impact on productivity and economic growth has not fully materialized. Realizing the full potential of Generative AI requires addressing implementation challenges, investing in complementary skills and infrastructure, and navigating ethical and societal concerns. As we move forward, a balanced perspective that recognizes both the capabilities and constraints of Generative AI will be essential for successful implementations while the field is still maturing.