# Is artificial intelligence leading to a new technological paradigm?

- **DOI:** [10.1016/j.strueco.2024.12.006](https://doi.org/10.1016/j.strueco.2024.12.006)
- **Rating:** Limited (2.8 out of 5)
- **Verdict:** Applicant-level patent evidence documents AI's widening industrial reach, but its claimed acceleration of further innovation lacks a counterfactual.
- **Framework version:** v5.10
- **Pipeline version:** v0.4
- **Evaluation date:** 2026-11-06
- **Evaluation page:** https://www.nabu.science/paper/10.1016%2Fj.strueco.2024.12.006

## Key claims

1. AI technologies may indeed generate a paradigmatic shift or, at least, a major radical transformation within the ICT paradigm
   - Support: Tentative
   - Type: Primary finding
   - Basis: unbenchmarked historical analogy cannot distinguish a new paradigm from transformation within ICT
   - Location: Abstract (final sentence); Section 5, Conclusions
2. AI patenting has accelerated and substantially evolved in terms of its pervasiveness, with AI innovators shifting from the ICT core industries to non-ICT service industries over the investigated period.
   - Support: Supported
   - Type: Secondary
   - Basis: descriptive sector-share trends support the shift within a keyword-defined, coverage-limited patent sample
   - Location: Abstract; Section 4.1 and Table 1
3. Moreover, there has been a decrease in concentration of innovation activities and a reshuffling in the innovative hierarchies, with innovative entries and young and smaller applicants driving this change.
   - Support: Tentative
   - Type: Secondary
   - Basis: descriptive co-movement does not identify entrants or younger applicants as drivers of the concentration change
   - Location: Abstract; Section 4.2-4.3, Figs 4-8
4. Finally, we find that AI technologies play a role in generating and accelerating further innovations (so revealing to be “enabling technologies”, a distinctive feature of GPTs).
   - Support: Tentative
   - Type: Secondary
   - Basis: before-and-after patent counts without a counterfactual cannot identify AI's contribution to subsequent innovation
   - Location: Abstract (final findings sentence); Section 4.4 and Table 2

## Strengths

- **Applicant-level evidence connects multiple innovation patterns** The matched patent–applicant dataset links industry composition, concentration, entry, and subsequent patenting for 23,915 AI innovators. This makes the paradigm argument empirically inspectable across multiple indicators.
  - Dimension: contribution
  - Location: Section 3; Table 1; Sections 4.1–4.4
- **Documented construction supports descriptive trend analysis** DOCDB simple families reduce duplicate invention counts, the keyword list is disclosed, and concentration is examined through multiple measures. The entry analysis also uses a six-year observation buffer.
  - Dimension: methodological rigour
  - Location: Section 3; Sections 4.2–4.3; Table A1
- **Competing interpretations receive substantive contextual treatment** The discussion engages both continuation and new-paradigm accounts. The conclusion connects early-stage sampling and later capital intensity to the difference between its concentration findings and concerns about recent AI market power.
  - Dimension: positioning
  - Location: Section 2; Section 5

## Limitations

- **Innovation acceleration lacks an identifying counterfactual** The before-and-after patent comparison does not separate AI's contribution from firm growth, selection, or existing innovation trajectories. Its 78.7% increase therefore does not establish the innovation-generating role asserted in the Abstract.
  - Dimension: methodological rigour
  - Location: Abstract; Section 4.4; Table 2
- **Measurement and coverage effects remain untested** The broad keyword list has no reported precision validation, and changes in country or institutional composition are not tested as explanations for the trends. Employment imputation further limits interpretation of the smaller-applicant result.
  - Dimension: methodological rigour
  - Location: Section 3; Section 4.3; Tables A1–A2; Fig. A3 note
- **Paradigm interpretation lacks a historical benchmark** The indicators match stylized features of earlier technological transitions but are not computed for a historical comparator. The observed patterns therefore do not distinguish a new paradigm from substantial change within ICT.
  - Dimension: contribution
  - Location: Section 2; Sections 4–5

## Reliability checks

Caveats (4 of 4 checks).

- **Results check:** Noted. 2 reported values do not reconcile, but the key claims do not rest on them.
  - an average three-year increase of just 5.9 % in the overall non-AI patenting activity within our full sample during the 2003-2013 period: "an average three-year increase of just 5.9 % in the overall non-AI patenting activity within our full sample during the 2003-2013 period" (Section 4.4, AI as an enabling technology) vs "On average, each AI innovator filed 24.4 non-AI patent families in the three years preceding their first AI patent and 43.6 in the three years following it." (Section 4.4, AI as an enabling technology). Alternative reconstructed benchmark definitions give approximately 11.7% for the 2003–2005 versus 2011–2013 windows or 5.8% for a three-year compounded growth calculation, so the unstated construction prevents determining whether 5.9% is erroneous. → Effect on result unclear
  - Minor reporting notes (1):
    - Descriptive: "% change between 2000-2005 and 2011-2016" (Table 1, last row, per-entity column) vs "the number of AI patent families every AI innovator applied for declined on average by 25 %" (Section 3, Data, sample and methods (sub-period comparison paragraph)). Using the yearly-row sums cited in the reviews, (62,576/23,810)/(16,636/5,433) implies approximately a 14.2% decline, rather than Table 1's 24.6%, while preserving the declining direction. → Conclusion holds
- **Reporting and transparency:** Noted. Figure notes use different applicant populations without a complete reconciliation. The employment analysis discloses substantial nearest-in-time imputation in the appendix but does not explain it alongside the main-text size result.
  - Data availability: not stated
- **Reference integrity:** Noted. 71 of 76 checkable references verified
  - 94 references in manuscript
  - 18 are books, websites or datasets — counted, but not index-checkable
  - Could not be verified automatically — queued for manual review:
    - Brynjolfsson, E., McAfee, A., 2011. Race against the machine: How the digital revolution is accelerating innovation, driving productivity, and irreversibly transforming employment and the economy. Digital Frontier Press, Lexington, Massachusetts. (Crossref bibliographic top candidate ratio 0.37 — no plausible match)
    - Dosi, G., 1982. Technological Paradigms and Technological Trajectories: A Suggested Interpretation of the Determinants and Direction of Technical Change. Research Policy 11 (3), 147–162. (Ambiguous: title ratio 0.57 via crossref_bibliographic sits between the divergence (0.30) and match (0.70) thresholds; too close to call either way, so it is recorded as unresolved rather than judged.)
    - Hotte, K., Tarannum, T., Verendel, V., & Bennett, L. (2024). Exploring Artificial Intelligence as a General Purpose Technology with Patent Data. A Systematic Comparison of Four Classification Approaches. arXiv preprint arXiv:2204.10304. (Ambiguous: title ratio 0.51 via datacite sits between the divergence (0.30) and match (0.70) thresholds; too close to call either way, so it is recorded as unresolved rather than judged.)
    - Schumpeter, J.A., 1912. The Theory of Economic Development. Harvard University Press, Cambridge (Mass.) ed., 1968. (Crossref bibliographic top candidate ratio 0.26 — no plausible match)
    - Schwab, K., 2017. The Fourth Industrial Revolution. New York, Crown Business. (Ambiguous: title ratio 0.70 via crossref_bibliographic sits between the divergence (0.30) and match (0.70) thresholds; too close to call either way, so it is recorded as unresolved rather than judged.)
  - References not verified automatically are resolved manually within 24 hours.

## Post-publication record

- **Status:** Clean
- **Summary:** No retraction or expression of concern found in Crossref (includes Retraction Watch).
- **Last checked:** 2026-11-06
