Can Artificial Intelligence Accelerate Technological Progress? Researchers' Perspectives on AI in Manufacturing and Materials Science

summary

The gist

The gist Artificial intelligence (AI) was primarily used for modeling of materials and manufacturing processes, facilitating cheaper and more rapid search of design spaces for materials and

In short

Researchers interviewed manufacturing and materials science experts found AI is primarily used for modeling materials and processes, speeding up design space searches and saving time, cost, and computation. While useful for existing data interpolation, AI lacks the ability to develop new scientific theory. The study suggests AI/ML supports sustaining innovation within current methods but requires conventional research to drive major breakthroughs.

Key concepts

Surrogate Models: These are AI models trained on existing physics-based simulations. They aim to produce similar results much faster and at a lower computational cost than running the original, complex physical models. This allows researchers to explore many more design options quickly without needing massive computing power for every test.
Phenomenological Modeling: This involves training AI models on input/output data when robust theoretical models are missing for a specific phenomenon. It helps create predictive tools where traditional physics-based methods cannot be applied directly, allowing researchers to address previously unsolvable problems by finding hidden patterns in the data.
Interpolation vs. Extrapolation: AI excels at 'interpolation,' meaning it can accurately predict outcomes within the range of data it was trained on. However, it struggles with 'extrapolation'—predicting results outside that known data space. This limitation means AI is best for refining existing knowledge rather than discovering entirely new, disruptive scientific theories.
Sustaining Innovation: The paper suggests AI/ML is a tool for 'sustaining innovation,' meaning it helps maintain and improve current technological processes efficiently. It acts as an aid within established paradigms, but it is not currently seen as a driver for fundamentally new, disruptive theoretical advancements.

Terminology used across episodes

Reinforcement Learning (RL) — explained in 25 episodes
Retrieval-Augmented Generation (RAG) — explained in 21 episodes
Sample Complexity — explained in 20 episodes
Flow Matching — explained in 19 episodes
Quantum Fisher Information (QFI) — explained in 19 episodes
Federated Learning — explained in 17 episodes

This episode discusses

The paper

Can Artificial Intelligence Accelerate Technological Progress? Researchers' Perspectives on AI in Manufacturing and Materials Science · Read on arXiv

John P Nelson, Olajide Olugbade, Philip Shapira, Justin Biddle

Oregon State University · Georgia Institute of Technology · Manchester Institute of Innovation Research Institute for Advanced Technology

Artificial intelligence (AI) raises expectations of substantial increases in rates of technological progress, but such anticipations are often not connected to detailed ground-level studies of AI use in innovation processes. Accordingly, it remains unclear how and to what extent AI can accelerate innovation. To help to fill this gap, we explore and assess results from 32 interviews with U.S.-based academic manufacturing and materials sciences researchers experienced with AI and machine learning (ML) techniques. We found that AI was primarily used for modeling of materials and manufacturing processes, facilitating cheaper and more rapid search of design spaces for materials and manufacturing processes alike. Benefits included cost, time, and computation savings in technology development. However, AI/ML tools were unreliable outside design spaces for which dense data were already available; they required skilled and judicious application in tandem with older research techniques; and concerns were raised about the potential to detrimentally circumvent opportunities for disruptive theoretical advancement. Based on these results, we suggest there is reason for optimism about acceleration in sustaining innovations through the use of AI/ML; but that support for conventional empirical, computational, and theoretical research is required to maintain the likelihood of further disruptive advances in manufacturing and materials.

DOI: 10.1016/j.respol.2026.105598

Transcript

Introduction to the show: ident: AI Radio. Generated commentary on the latest Artificial Intelligence papers.

Tom: Hi, I'm Tom.

Jane: And I'm Jane, here with Lu, Meng and our in-house large language model, Lalam.

Tom: We're looking at the paper: "Can Artificial Intelligence Accelerate Technological Progress? Researchers' Perspectives on AI in Manufacturing and Materials Science".

Jane: Here's the gist of the paper. The gist Artificial intelligence (AI) was primarily used for modeling of materials and manufacturing processes, facilitating cheaper and more rapid search of design spaces for materials and manufacturing processes alike, with benefits including cost, time, and computation savings in technology development.

Tom: How it works The research explored how AI can accelerate technological progress by investigating whether it improves the breadth, reliability, or cost at which new technological alternatives can be tested and whether it is taking over research tasks previously performed by humans.

Jane: The core question of whether and how AI can facilitate technological progress has not been answered, requiring an investigation into how AI applications and affordances affect fundamental components of the process of technological change.

Tom: The study framed the question within longstanding evolutionary accounts of technological change, asking whether AI improves the breadth, reliability, or cost at which new technological alternatives can be tested and whether it is taking over research tasks previously performed by humans.

Paper summary: Tom: We're wrapping up our look at "Can Artificial Intelligence Accelerate Technological Progress? Researchers' Perspectives on AI in Manufacturing and Materials Science." It really boils down to whether this technology is accelerating things or just helping us do the same work faster.

Jane: It really gets to that big question again, right? Can AI actually speed up progress, or is it just a fancy modeling tool we use for existing work?

Tom: The authors did a survey of thirty-two researchers and found that AI was primarily used for two main things: building models of materials and manufacturing processes.

Lu: They focused on finding new design spaces for those materials, which made the search process faster and cheaper.

Jane: But there were big caveats about when this AI actually works well outside of those specific data-dense design spaces.

Tom: Right, they found that outside of those dense areas, the tools weren't reliable at all because they were limited to interpolation instead of extrapolation.

Meng: And there’s this scientific concern that AI doesn’t have a way to develop genuinely new scientific theories itself, it’s stuck in what it was trained on.

Jane: So the core idea here is that AI is more of a tool for sustaining existing innovation within current technological paths, not necessarily inventing the next big thing on its own.

Tom: Exactly. They suggest that AI/ML is best used as another rung to test different material or manufacturing ideas more cheaply, but it’s not replacing the need for conventional experimental and theoretical research methods.

Lu: The implication for researchers is that they still need to focus on building robust systems of traditional theory development alongside these AI tools.

Jane: So if you're tracking this, you have to watch how funding supports both the AI infrastructure and those older, more traditional research skills.

Conclusion: Tom: So we've been looking at those interviews from manufacturing and materials science researchers about how AI fits into innovation processes, and now we’re going to wrap up with some thoughts on what that actually means for the bigger picture.

Jane: It really boils down to whether this technology is genuinely accelerating progress or just helping people do the same work faster within established limits.

Tom: The paper by Author one and Author two, titled "Can Artificial Intelligence Accelerate Technological Progress? Researchers' Perspectives on AI in Manufacturing and Materials Science," surveyed thirty-two academics.

Lu: They found that for the most part, the AI was focused on modeling materials and manufacturing processes.

Jane: Specifically, they saw it helping researchers search design spaces for new materials and processes much faster and cheaper than before.

Tom: And the benefits they highlighted were all about cutting down costs, saving time, and reducing the computation needed for technology development.

Meng: But there’s this big catch: those AI tools only work reliably in areas where they already have a lot of good data from before.

Jane: Right, because outside those dense data sets, the AI just gets unreliable when it tries to guess things that are totally new territory.

Tom: That means the models can't predict totally unknown areas without some human guidance involved in how they use them.

Lu: And a major scientific worry they brought up is that the AI itself isn't good at creating entirely new scientific theories; it’s primarily focused on fitting what it already knows.

Jane: So, the core idea here is that AI seems better suited for sustaining existing innovations within current technological paths rather than inventing brand new theoretical concepts from scratch.

Tom: They suggest that AI/ML is really just another tool in the research toolkit, helping you test different material ideas more cheaply than before.

Lu: The implication for those researchers is that they still need to keep building those conventional experimental and theory development methods as the main backbone of their work.

Jane: It sounds like they're saying AI fits into the workshop, but you can't just replace the whole workshop with just one tool.

Tom: Exactly, it should be seen as adding another rung to a ladder of trial rather than completely replacing deep empirical work.

Meng: They also stressed that you need skilled people applying these tools alongside traditional research skills to actually make them useful in a real setting.

Lu: The focus needs to stay on building that whole system around the AI, not just trying to get the models themselves perfectly tuned.

Jane: So if you're tracking this, keep an eye on how funding supports both the AI infrastructure and those older, more traditional research methods moving forward.

More episodes

← Home