Economists of the Federal Reserve are not sold that AI will make workers more productive, saying that this could be a unique invention like the bulb

Entitled “AI generative at the crossroads: bulb, dynamo or microscope?” The article, written by Martin Neil Baily, David M. Byrne, Aidan T. Kane and Paul E. Soto, explores if the AI generation represents an ephemeral innovation or a revolutionary force similar to general technologies (GPT) such as electricity and internet.

The Fed economists finally conclude that their “modal forecasting is for a significant contribution from the AI generation to the level of labor productivity”, but careful that they see a wide range of plausible results, both in terms of total contribution to make workers more productive and how speed could happen. To return to the metaphor of the light bulb, they write that “certain inventions, such as the bulb, temporarily increases the growth of productivity as adoption spreads, but the effect fades when the market is saturated;

Here is why they consider it an open question if Gen Ai can end up being a sophisticated technological version of the bulb.

Gen Ai: a tool and a catalyst

According to the authors, General AI combines the features of GPTs – those that trigger innovation stunts in the sectors and continue to improve over time – with the characteristics of “inventions of invention methods” (Imis), which make research and development more effective. The authors see the potential for Gen AI to be a GPT like the Electric Dynamo, which has continually triggered new commercial and efficiency models, or an IMI as the compound microscope, which has revolutionized scientific discovery.

FED economists have warned of the start of technology development, writing: “The case that generating AI is a convincing technology is convincing, supported by the impressive record for training and basic innovation innovation.”

Since Openai launched Chatgpt at the end of 2022, the authors said that Gen AI had demonstrated remarkable capacities, the correspondence of human performance on complex tasks for the transformation of first line work in writing, coding and customer service. That said, the authors noted that they found little evidence that many companies really use technology.

Limited but increasing adoption

Despite such a promise, the document stresses that most of the gains have so far been concentrated in large companies and digital native industries. The surveys indicate a large -scale adoption among large companies and the technology -oriented sectors, while its use in small businesses and other functions are lagging behind. The data of job offers only show a modest growth in the demand for skills in AI since 2017.

“The main obstacle is dissemination,” write the authors, referring to the process by which a new technology is integrated into generalized use. They note that the typical productivity of GPT’s handicaps such as computers and electricity took decades to take place while restructured, invested and developed companies.

“The share of jobs requiring AI skills is low and has only increased modestly, which suggests that companies adopt a cautious approach,” they write. “The ultimate test of the question of whether the AI generation is a GPT will be the profitability of the use of the AI generation on a large scale in a commercial environment, and these stories are difficult to find at present.” They know that many people use technology, “perhaps without their employers”, and they speculate that the future use of technology can become so routine and “banal” that companies and workers no longer know how used it is.

Knock-on and complementary technologies

The report details how the AI generation already stimulates a wave of innovation of products and processes. In health care, AI tools write medical notes and help radiology. Financial companies use GEN AI for compliance, subscription and portfolio management. The energy sector uses it to optimize network operations, and information technology see several uses, programmers using Github Copilot to perform 56% tasks more quickly. Call centers operators using conversational AI have also seen an increase in productivity by 14%.

Meanwhile, the progress in progress of the equipment, including rapid improvements in chips called graphic treatment units, or GPU, suggest that the underlying engine of Gen Ai still accelerates. The patent deposits linked to AI technologies have increased since 2018 since 2018, coinciding with the rise of the architecture of the transformers – a backbone of the important language models of today.

“ Green Shots ” in research and development

The article also finds the generation AI acting more and more as an IMI, improving observation, analysis, communication and organization in scientific research. Scientists now use GEND to analyze data, write research articles and even automate parts of the discovery process, although questions remain on the quality and originality of the exit generated by AI.

The authors highlight increasing AI references in R&D initiatives, both in patent data and corporate profit calls, such as additional proof that the IA generation takes a foothold in the innovation ecosystem.

Cautious optimism and open questions

Although the prospects for an increase in productivity focused on the Gen-Ai are promising, the authors warn against waiting for the transformation overnight. The process will require significant additional investments, organizational change and reliable access to computer and electric energy infrastructure. They also highlight the risks of investing blindly in speculative trends – a lesson in past technological booms.

“The contribution of generation AA to productivity growth will depend on the speed at which this level is reached, and historically, the process of integration of revolutionary technologies in the economy is prolonged,” concludes the report. Despite these uncertainties, the authors believe that the double role of the AI generation – as a transformative platform and as a method for accelerating invention – well for long -term economic growth if obstacles to general adoption can be overcome.

Yet, and if not another bulb?

For this story, Fortune Used a generative AI to help an initial project. An editor checked the accuracy of the information before the publication.