Recent Research: How AI is changing the nonprofit research institute

While some college computer engineering profs may be advising their students not to worry about artificial intelligence derailing their salary projections and long-term career options, it would appear businesses are getting on with deploying the latest AI advances as quickly as possible to see what improvements might be made for the firms’ productivity rates and bottom lines. A recently released working paper from Germany’s Fraunhofer Institute for Systems and Innovation Research (Fraunhofer ISI) reports on an early analysis of AI adoption in the innovation research process. The authors’ preliminary conclusion?  “AI is currently used more for competitive differentiation, but in 15 years it can become the standard as a so-called basic technology and thus a factor critical to competition.”

The authors suggest, however, that “the energy-intensive use of AI will remain an immense challenge.” One may wonder, in addition to the authors’ observations, if AI electricity demand will drive electricity rates to the point that only the companies in the best market and financial position will be in a position to optimize AI use, particularly when AI energy draw is coupled with other pressures on electricity generation such as the transition to electric vehicles, de-gasification of kitchen stoves, ranges and water heaters, other smart home appliances, and cryptocurrency.

Founded in 1972, employing approximately 300 staff currently, and conducting roughly 400 research projects annually on a budget of 33.3 million euros (in 2022), the Fraunhofer ISI was used as the case study for the investigation. Authors Malte Bush and Daniel Duwe conducted interviews of members of six Fraunhofer ISI departments, asking how the staff were already using AI in the innovation process, challenges they had encountered as AI has been adopted, and paths for more fully exploiting AI’s potential value at the research institution. In their efforts, the authors identified the opportunity within the institute for better inter-departmental coordination across AI technology use and learning.

Reading like an internal report to inform the institute’s—or any nonprofit research center’s—future business strategy, the first third of the paper provides a useful summary of their literature review on previous research into how AI is impacting the innovation process. Of particular value and highlighting the pace at which AI is being integrated into R&D and innovation, nearly all of the previous papers described are four years old or less.

The interviews yielded comments on the staff members’ perceptions of AI (positive opportunity for efficiency gains; highly variable in strength and outcomes; and there was bias, doubt or ideological determinism in potential/perceived value) and the level of understanding the difference between digitalization and AI’s learning characteristic. A recommendation for readers considering AI’s role in their own organizations might be to ensure a basic understanding among employees of what AI is, is not, and may be. Setting some minds at ease may be beneficial as well regarding the potential implications for future employment through the introduction of AI to the organization’s innovation process and other workflows.

AI may permeate—to the organization’s advantage—all aspects of the innovation process, the researchers found through their interviews. “Great added value” was found by many in the ideation stage and sorting data and historical research for determining novelty and practicality of ideas. AI was also found to be useful in the development and market launch phases of new products in assessing effective business models and customer identification/sales channel strategy/messaging, respectively.

Another dimension of Fraunhofer ISI’s findings looked at the varied concrete effects AI might have based on industry sector. For their particular portfolio, the impacts from AI ranged from significant ideation value (in pharmaceuticals), data efficiencies (in science and services), application-oriented (in energy and smart cities), or transformative business models and efficiencies (innovation systems and general management). Examples across all stages of the innovation-to-market process and by sector are provided in a succinct summary matrix.

Obstacles identified by the Fraunhofer ISI team included”:

• Keeping up with employee qualifications, competencies, pricy wage demands, and skills retraining requirements in a rapidly changing field;
• Ethics and copyright infringement issues surrounding AI;
• Data sourcing, completeness, accuracy and currency;
• Reliability, inherent biases, and quality of AI outcomes;
• AI’s large carbon footprint and negative environmental impacts (for instance, through high energy demand and corresponding emissions, already mentioned); and,
• Evolving application regulatory barriers.

The authors conclude that AI has multiple roles to play throughout the innovation process and that the innovation process itself will evolve as AI is adopted and itself evolves. What that means for nonprofit research centers remains to be seen, but sitting on the sidelines waiting for your organization’s future to be determined for your organization may not be in its best interest over the long term.  Whether you have 15 years or less is also unknown.

The working paper, “Artificial intelligence in innovation processes. A study using the example of an innvation (sic) research institute” is available here.