The Impact of Generative AI.
Chapter 1: The Global Economy and the Job Market
Generative artificial intelligence (GAI) has been making headlines with the rapid popularization of ChatGPT to the top of the app charts. The pace of adoption is something we have not seen before. As users of GPT-4 internally, we see it as a promising technology that can drive significant productivity gains, especially in repetitive tasks. As investors, we contemplate the significant changes that broad adoption of this technology could bring to the world, starting with what we see as the impact on the global economy and the job market.
What makes generative AI particularly fascinating is its ability to generate content that closely resembles that produced by humans. Hence, we believe these tools will usher in significant changes in global industries, with AI streamlining workflows and automating tasks. Within our own team, we use GPT-4 and we are already witnessing small victories that are driving real productivity improvements in our daily lives. Specifically, we use GPT-4 to summarize sell-side research and industry channel checks, which are then tracked in a table that ChatGPT maintains for us.
To benchmark the productivity gains with actual data, we look at research published by MIT and Stanford. Their paper suggests that the introduction of a generative AI tool in a customer support setting boosted agent productivity by 14% on average, by further automating the process of solving customers’ requests and complaints. The tool particularly benefited novice and lower-skilled workers who experienced a 35% increase in productivity.
With the anticipated launch of Microsoft Co-Pilot within Office 365 later this year, we expect knowledge workers worldwide to start benefiting from GPT-4 in everyday tasks in tools like Outlook, Word, Excel, and PowerPoint. Microsoft is already profiling the technology where it can create presentations, spreadsheets, and author documents for you. With more than 345 million paying Office 365 users worldwide, any productivity increase is set to have broad, but more importantly tangible, implications globally.
Impact On Jobs and Employment
While we believe these productivity gains will benefit workers, it is unclear to whom the added value of the technology will accrue in the intermediate term. For instance, Goldman Sachs estimates that two-thirds of US jobs are exposed to AI automation to some degree. However, among these jobs, some have a low percentage of tasks that can be automated, while others have a significant portion of their workload at risk. Goldman Sachs points out that administrative and legal professions could automate almost half of their tasks with generative AI. In contrast, manual work like construction, installation, and maintenance are much less exposed.
If AI renders many workers far more productive, we could initially see a broad slowdown in hiring, which might eventually lead to some job cuts. To quantify this thought, Barclays estimates that AI agents in contact centers could decrease the total number of agents from 17.5 million agents today to 9 million by 2032, a 49% decrease in total agent count. Replacing this agent growth with more interactions being handled by AI could lead to 43% in total cost savings, still according to Barclays.
In the near to intermediate term, this could have negative impacts to wage growth and potentially the employment rate. Wage growth is a significant factor in services inflation, which the Federal Reserve has been focusing on during the current rate hike cycle.
With potential layoffs on the table due to these gains in productivity, we recall that in 2022, 1,056 tech companies laid off 165,000 employees, according to layoffs.fyi. Following these mass layoffs, the Sycomore team engaged with both portfolio companies and others to better understand the motivations and the ways the layoffs were conducted. We engaged with Zoom, Gitlab, Accenture, Salesforce, Microsoft, SAP and Amdocs on their respective layoff programs to make sure these programs were announced and managed in an ethical way.
Impact on Macro
Conversely, increases in productivity and efficiency should generally propel economic growth, as the labor hours (cost) saved relative to increasing output (revenues) can allow companies to invest in other more productive or strategic inputs.
According to Goldman Sachs, the labor productivity booms following previous significant technological advancements (electricity, and personal computing) led to 1.5% gains in labor productivity per year over an ensuing period of ten years in the US. These productivity booms coincide more with mass market penetration of the technological advancements, rather than the actual commercial availability. Goldman points to the ~50% adoption mark as the approximate point where the productivity boom begins.
Reaching the halfway 50% mark took 20 years for electricity and the personal computer. However, it’s been faster for newer technologies like the tablet market, which took about 5 years to go from 0% to 50% adoption.
This means it can be used through apps or web browser on PCs and mobile phones that enterprises and consumers already own and know how to use. In our view, the typical frictions to adoption (baring regulatory issues) should be minimal. This is already evident with ChatGPT reaching 100mn monthly users in 2 months; while Tik Tok, Instagram, Spotify, and Uber, for example took 9, 30, 55, 70 months, respectively.
Goldman estimates that the tailwinds to economic growth from AI globally could increase GDP by 7% over ten years. However, given the high velocity of adoption already evident, and the broad applicability of the technology across industries, we ponder whether the impact could be even bigger, and faster.
Impact on Public Policy
If generative AI impacts the economy and our professional lives as much as we anticipate, we also expect an inevitable reshaping of social policies. As automation increases, it's likely that policymakers will need to reassess social safety nets, particularly for those in occupations at high risk of automation. Job retraining programs may become a focal point, helping to equip workers with the skills needed in the AI-driven economy. If AI instead has detrimental effects on human employment, at what point do concepts like universal basic income (UBI) come into play? Read “more government spending”.
However, the policy implications are not limited to employment and federal budgets. AI has the potential to significantly impact privacy and data security, necessitating a fresh look at existing regulations. We could also see malicious use of AI in political spheres on social media. Hence, policymakers will need to address the increasing threat of misinformation and its impact on democratic processes. A problem already pervasive in many countries that will only be amplified.
As with any technological advancement, the key will be in balancing the immense potential benefits of generative AI with the novel challenges it presents.
This short note serves as the first volume of a series where we intend to go deeper into the implications for different industries, such as healthcare, construction, and retail, as well as its influence on the various subsectors in technology. We hope to frame how generative AI could reshape our world in the years to come, through investing and sustainability lenses.