The September 16, 2023 issue of The Economist included a special focus on “How AI Can Revolutionize Science” with three articles on the topic. “Debate about artificial intelligence (AI) tends to focus on its potential dangers: algorithmic bias and discrimination, the mass destruction of jobs and even, some say, the extinction of humanity,” noted the issue’s lead article. “As some observers fret about these dystopian scenarios, however, others are focusing on the potential rewards. AI could, they claim, help humanity solve some of its biggest and thorniest problems. And, they say, AI will do this in a very specific way: by radically accelerating the pace of scientific discovery, especially in areas such as medicine, climate science and green technology.”
But, while AI-based innovation will lead to extraordinary progress, “it faces a largely hidden threat: Innovation is becoming harder and more expensive,” said “The next innovation revolution—powered by AI,” a June 2025 report by McKinsey’s AI unit Quantum Black, “By many metrics, and in many fields, each dollar spent on R&D has been buying less innovation over time,” the authors added, citing concrete evidence of the decline of R&D productivity in two key areas:
The semiconductor industry had an exponential growth rate of 35% in transistors between 1971 and 2014. But, during this same period, semiconductor companies and equipment manufactures estimate that their annual R&D expenditures rose by a factor of 18. “In other words, maintaining the performance growth rate in Moore’s Law required 18 times more inflation-adjusted R&D spending in 2014 than it did in 1971.”
The biopharmaceutical industry has produced innovative products used to prevent and treat many diseases, enabling millions of people to live longer and healthier lives. But due to declining R&D productivity, drug discovery has become slower and more expensive, as “the number of new drugs approved per billion US dollars spent on R&D halved roughly every nine years between 1950 and 2011, falling around 80-fold in inflation-adjusted terms.”
“AI has the potential to bend the curves of R&D productivity, not only unlocking more economic growth but also boosting the chances of solving some of the most important human challenges, from preventing and curing diseases to reducing the level of carbon emissions,” said the McKinsey report.
How can AI reignite innovation productivity? According to McKinsey, a simplified model of the R&D process consists of identifying a set of customer needs, generating candidate designs, and then evaluating those designs to identify the most promising ones that will best meet the customer needs. Based this simplified model, the report identifies three primary ways for AI technologies to reignite innovation productivity:
- increasing the velocity, volume, and variety of design candidate generation;
- accelerating the evaluation of candidates through AI proxy models; and
- accelerating research operations.
Let me say a few words about each.
Increasing the velocity, volume, and variety of design candidate generation
In 2020, OpenAI released GPT-3 its latest and most powerful foundation model. Shortly afterwards, its creators discovered that not only could GPT-3 generate whole sentences and paragraphs in English in a variety of styles, but it had developed surprising skills at writing computer software even though the training data was focused on the English language, not on examples of computer code. But, as it turned out, the vast amounts of Web pages used in its training included many examples of computer programming accompanied by descriptions of what the code was designed to do, thus enabling GPT-3 to teach itself how to program. GPT-3 could also generate legal documents, like licensing agreements or leases, as well documents in a variety of other fields.
But in addition, foundation models like GPT-3 can be trained to generate outputs other than human language, documents, and computer code. Properly trained, these AI models can be used to identify molecules with particular properties in drug discovery or materials with the characteristics needed in batteries and solar cells. One of the highest potential opportunities for AI to enhance innovation is to more quickly generate a greater volume and variety of design candidates.
For example, “How scientists are using artificial intelligence,” one of the articles in the aforementioned September 2023 issue of The Economist, described how AI helped find new antibiotics, salicin and abaucin, for use against two of the most dangerous known antibiotic-resistant bacteria. “In both cases, the researchers had used an artificial-intelligence (AI) model to search through millions of candidate compounds to identify those that would work best against each ‘superbug’. The model had been trained on the chemical structures of a few thousand known antibiotics and how well (or not) they had worked against the bugs in the lab. During this training the model had worked out links between chemical structures and success at damaging bacteria. Once the AI spat out its shortlist, the scientists tested them in the lab and identified their antibiotics.”
The same article discussed how AI helped accelerate the search of new materials for batteries. “When researchers at the University of Liverpool were looking for materials that would have the very specific properties required to build better batteries, they used an AI model known as an ‘autoencoder’ to search through all 200,000 of the known, stable crystalline compounds in the Inorganic Crystal Structure Database, the world’s largest such repository. The AI had previously learned the most important physical and chemical properties required for the new battery material to achieve its goals and applied those conditions to the search. It successfully reduced the pool of candidates for scientists to test in the lab from thousands to just five, saving time and money.”
Accelerating the evaluation of candidates through AI proxy models
The next step in the product development life cycle is to evaluate which of the various candidate designs are the most promising. Over the years, manufactures have developed a variety of methods to evaluate different product designs.
One common technique for physical products like cars is to build prototypes and then subject them to a variety of physical tests, like crash worthiness and aerodynamics analysis. But physical prototypes and tests can be costly and time consuming, especially when applied to a large number of design options.
Another method is to build mathematical models of the products being evaluated and then analyze different design options with powerful supercomputers. Such computational simulations have become more prominent given the major advances in lowering the price and increasing the price/performance of advanced parallel supercomputers.
“But a recent discovery found that it is possible to repurpose the neural network technology developed for AI systems to train models that can act as proxies for more computationally intensive physics-based models. These AI-style surrogate models do not imitate the thinking that people do; instead, they predict the outcomes of physical phenomena in the world. When used to predict the behavior of a complete system, these models are akin to a digital twin.” Such AI-based techniques are being used experimentally in weather forecasting, in the evaluation of different car and airplane designs, and the testing and evaluation of biological drugs.
Accelerating research operations
“In addition to generating and evaluating design candidates, there are several additional ways that LLMs, sometimes coupled with other AI technologies, are being used to accelerate various activities in the product development process”:
- Identifying and analyzing customer/user needs, products, and features. “LLM-powered software solutions are being used, particularly by consumer companies, to synthesize a vast array of product reviews, social media posts, customer service transcripts, and other sources of customer data to identify addressable market segments and the product categories and features/functions that would best address the as-yet unmet needs of customers.”
- Exploring and synthesizing existing research and data. “In industries such as life sciences, chemicals, and materials, there is a vast and rapidly growing body of published research and databases. It can be challenging for scientists to keep up with the literature in their own subdiscipline, not to mention the adjacent or even distant areas of other research, which could bring insights for breakthroughs in their field.”
- Streamlining internal knowledge management. Large corporations hold a huge amount of knowledge in various databases and tacit knowledge in the minds of employees. “LLM-powered tools can help to codify tacit knowledge — say, transcribing and capturing recorded meetings and other communications (with the permission of the participants, of course).”
- Automating documentation tasks. In highly regulated industries “such as pharmaceuticals and aircraft manufacturing, there are significant documentation requirements — for example, for regulatory filings, engineering change orders, and other required documentation. LLMs can accelerate the process of both generating and synthesizing these documents.”
- Collaborating with humans for ideation and concept development. Product managers, scientists, engineers, designers, and other participants in the product development process can “converse” with LLMs to stimulate ideas, get “opinions,” and have their ideas challenged, much as they would with a colleague. These experiences illustrate that it is possible for humans and AI to collaborate, but the human skill in using AI tools can significantly influence the effectiveness of these collaborations.”
What business leaders can do to harness the power of AI in R&D
Finally, the McKinsey report recommends that leaders consider four key levers for leveraging AI to accelerate innovation:
- Move quickly and scale rapidly. Climbing this learning curve sooner — and faster — can help you to gain a competitive edge over others.
- Rewire your organization beyond just tech. Beyond technology, capturing the value of AI requires aligning with the business strategy, building the right talent, agile adoption and scaling, and proper change management and governance.
- Build a core competency around models. A new critical core competency will be evaluating, integrating, training/adapting, and making build-versus-buy decisions about AI models, including open-source models, procured models, and even internally trained models, as part of the R&D process.
- Be thoughtful about incorporating humans in the loop. People will still have a major role in the R&D process, but those roles are likely to shift considerably in an AI-enabled future, requiring reskilling. Organizations will have to identify when it is critical to have a human in the loop, for example, to ensure safety or to sign off on various decisions where having an accountable individual is critical.”