LTO 2024 – Contribution on AI
Back
Article Type
Insights
Topic
Long-Term Outlook
Published
September 13, 2024
Read time
5 mins
Author
Igor Quint
Share

Despite the rapid technical advancements in artificial intelligence (AI) and generative AI, their impacts on the global economy seem subtle at first sight. While companies operating in the AI sector – especially those in the semiconductor industry – play a substantial role in driving global market indices, the direct impact of the technology on global GDP, inflation and the job market seems limited so far. In this section we dive deeper into current trends and explore how these advanced technologies may shape the economy going forward.

In last year’s long-term outlook, we proposed a two-stage illustration (see figure below) that linked the influence of AI within firms to their willingness to adopt AI and the ease of model deployment. Our presumption was that in the short term, most firms would fall under the “AI-augmentation” scenario, where modern AI tools would enhance workers’ capabilities, thereby increasing productivity, reducing human error and improving quality.  

Figure 1: Illustration of AI adoption pathways Source: Aegon Asset Management 

The “Full AI-integration” scenario would unfold when both the willingness to adopt and the ease of implementation for AI technologies were high. Factors such as novelty, lack of expertise and legislative uncertainty could deter most firms from adopting AI in the short run, while current technological capabilities and data safety and privacy concerns could impede implementation. This led us to the assumption that full AI integration (and its significant impact on the global economy) would only start to become widespread in the long term.

To evaluate the validity of this assumption against real-world developments one year ahead, let’s examine a few aspects in more detail. By assessing current legislative efforts, we can try to foresee whether they remove existing barriers or add to them. By determining the impact of current AI on the total productivity per worker, we can foresee whether AI will deliver on the expected GDP improvements. Lastly, through assessing current job markets, we can learn if and how AI is impacting the net number of occupations today.

Total labor productivity per worker

Research by Microsoft[1] indicates that the use of their large language model (LLM)-driven AI software saves surveyed users an average of 14 minutes daily due to productivity gains. Another study (Dell'Acqua et al., 2023)[2] found that among a sample of consultants, those with access to an LLM (GPT-4) completed more tasks on average (+12.2%), completed them in 25.1% less time, and had 40% higher quality results. Similarly, a study by the National Bureau of Economic Research (NBER)[3] found that the implementation of LLM-based assistants in customer support increased worker productivity by 14%, with the most substantial gains observed among less experienced workers. Assuming these findings carry over to more sectors and effects may even be magnified with increased technological progress, total factor productivity (TFP) is bound to grow. In a classical Cobb-Douglas form production function, an increase in TFP leads to an increase in total output (GDP), given constant capital and labor input. As such, AI is likely to contribute positively to the GDP growth rate, with larger effects expected when technology progresses, and adoption becomes more widespread.

Legislation

New regulations around the use and development of safe and responsible AI are being established in most major markets (Christodorescu et al., 2024)[4]. Frontrunning these efforts is the European Union with their Artificial Intelligence Act, effective from August 1st, 2024. It imposes requirements on AI applications, including transparency for “General-purpose AI” like large language models (LLMs).

While there is a grace period before the law applies to existing models, the implications of the regulation seem substantial. AI systems, especially those that are complex and powerful, are mostly “black boxes”. Considering the steep development costs, firms might hesitate to reveal their models and training weights fully, which could limit the presence of their models in the EU market. Consequently, the inability to access potent AI models could potentially impact the competitiveness of firms based in the EU.

At the same time, the US is advancing in AI legislation, highlighted by a 2023 executive order. The US focuses on guiding government AI use and promoting AI R&D, with an emphasis on ethical use and data privacy. Some state-level efforts implement safeguards or bans on high-risk governmental AI applications. These legislative efforts could shape corporations’ AI adoption and implementation in both regions. The EU’s stringent regulations might deter AI adoption due to transparency requirements which could limit access to powerful AI models, while the US’ focus on ethics and privacy could encourage adoption, despite potential state-level challenges. The rapidly evolving landscape requires ongoing observation to understand trends and impacts.

Job market: software developers

We previously argued that within the full AI-integration scenario (long term) we expect job redundancy to outpace newly created occupations, while the effects on the short term are likely not too pronounced. It is incredibly hard to quantify the current effects of artificial intelligence on job creation and replacement overall. To better understand the dynamics of AI in the workplace, let’s consider the field of software development.

In a survey among 500 US-based software developers by code repository GitHub, 92% of participants are already using AI coding tools both in and outside of work. 70% of developers say they see AI offering them an advantage in terms of coding quality, completion time and resolving incidents (Github, 2024)[5]

A 2024 study on developers’ perceptions of the impact of ChatGPT in software development (Vaillant et al., 2024)[6] provides insights into the expected impact of (generative) AI in the field of software development. Approximately 43% of the surveyed participants anticipate potential layoffs induced by the application of AI technologies in the workspace. 48% of participants are unsure or anticipate little impact, with just 8% of respondents not expecting any impact at all.

While advances in AI research and increasing implementation might (initially) involve a higher demand for developers and computer scientists, current evidence of this effect is absent. In fact, AI-related positions relative to total job listings decreased[7] in 2023, a trend also attributed to broader layoffs in the tech sector. These larger layoffs have significantly reduced the number of software development listings in the US[8], a phenomenon that has been present longer than the current entry of generative AI on the world stage. Yet, the advancements of generative AI are likely to threaten future job prospects[9] in the software development sector, with a few clear examples already occurring[10].

Current implications

The rapid advancements in AI technology present a complex picture for the global economy. While AI’s direct impact on macroeconomic indicators like GDP, inflation and job markets is hard to gauge, its potential is undeniable. Legislative efforts in major markets, such as the EU’s AI Act and the US’ executive order, are gaining traction. These efforts may hinder AI adoption and competitiveness but also promote clarity by removing ambiguity. In the job market, particularly in software development, AI is already making its presence felt, though its broader implications are still unfolding.  It’s potential positive impact on economy-wide productivity will likely take years to fully materialize. In the short run, financial markets will remain focused on AI due to its massive impact on current and future corporate profitability in the semiconductor and tech sectors. It will also be interesting to see whether AI can disrupt business models within and beyond the tech sector.

As AI continues to evolve, we keep monitoring the space and remain alert to broader implications.

This article is a preview of AAM’s Long Term Outlook for 2025-2028. The comprehensive outlook, featuring macroeconomic scenarios, expected returns, and additional articles, will be available in October.

[1] What can Copilot’s earliest users teach us about AI at work? (n.d.).

[2] Dell'Acqua, F., McFowland, E., Mollick, E. R., Lifshitz-Assaf, H., Kellogg, K., Rajendran, S., ... & Lakhani, K. R. (2023). Navigating the jagged technological frontier: Field experimental evidence of the effects of AI on knowledge worker productivity and quality. Harvard Business School Technology & Operations Mgt. Unit Working Paper, (24-013).

[3] Brynjolfsson, E., Li, D., & Raymond, L. R. (2023). Generative AI at work (No. w31161). National Bureau of Economic Research.

[4] Christodorescu, M., Craven, R., Feizi, S., Gong, N., Hoffmann, M., Jha, S., ... & Turek, M. (2024). Securing the Future of GenAI: Policy and Technology. Cryptology ePrint Archive.

[5] Shani, I. (2024, February 8). Survey reveals AI’s impact on the developer experience - The GitHub Blog. The GitHub Blog.

[6] Vaillant, T. S., de Almeida, F. D., Neto, P. A., Gao, C., Bosch, J., & de Almeida, E. S. (2024). Developers' Perceptions on the Impact of ChatGPT in Software Development: A Survey. arXiv preprint arXiv:2405.12195.

[7] Perrault, R., & Clark, J. (2024). Artificial Intelligence Index Report 2024.

[8] Schneider, M. (2024, April 2). Indeed’s 2024 US Jobs & Hiring Trends report. Indeed Hiring Lab.

[9]  Korducki, K. M. (2023, October 3). Computer science is no longer the safe major. The Atlantic.

[10] Thorbecke, C. (2023, July, 4). AI is already linked to layoffs in the industry that created it. CNN.  

Disclaimer

Author

Igor Quint
Investment Strategist
Igor Quint joined Aegon Asset Management in 2019 and works as an Investment Strategist in the Fiduciary Services and Investment Solutions team. 
View profile
Related Articles
View all Aegon insights