The threat of being disrupted is prompting companies to invest heavily into AI. How do they actually go about about evaluating, approving, and scaling these investments?
In this episode of Work Matters, Thomas Bertels talks with Dennis Glacken, industry professor at the Stevens School of Business. Dennis recently interviewed 32 senior IT leaders across 30 companies about how they select, govern and scale their AI investments. He expected to find a meaningful share of budgets set aside for disruptive bets. He found the opposite: most AI spending runs through the same process as any other IT project, and most of it targets personal productivity.
Dennis points out that most companies approach AI similar to how Lean Six Sigma was deployed - train lots of people, encourage lots of projects, focus on personal productivity. He argues that unless the focus shifts to salient, high value projects, that approach will deliver real but only marginal gains.
Another interesting finding was that most companies, AI governance required an additional layer of bureaucracy. Dennis explains why every project becomes a zoning variance, why traditional IT governance models don't work for AI, and what is needed to accelerate.
Also in this conversation:
- AI-native development vs. bolting AI onto legacy systems
- what fast in-house building means for SaaS pricing
- the data leakage problem when only some employees get licenses, and
- who in the C-suite should own the transformational bets if the CIO can't
Getting an AI project started is easy. Realizing real business value is hard.
Guest Bio:
Dennis Glacken is an Industry Professor at Stevens Institute of Technology, teaching Digital and Process Innovation in the business school’s undergraduate and graduate programs. Dennis brings more than 30 years in senior management roles at leading companies. His background includes leading IT, process innovation, and acquisitions, with a focus on portfolio management and aligning technology strategy with business objectives.
https://www.linkedin.com/in/dennisglacken/
Chapters:
0:00 Everyone's racing to do AI. Few are racing to disrupt.
2:24 Expecting disruptive bets, finding incremental ones
3:04 AI projects: easy to start, hard to put into production
4:54 Train everyone and let the ideas bubble up
6:31 Rolling out AI like Lean Six Sigma — and the backlash risk
7:03 The productivity paradox
10:24 The missing middle: no forum for cross-functional trade-offs
11:42 AI governance - why every AI project is a zoning variance
17:01 The real question is: "Should AI do this?"
20:29 Tens of thousands of lines of code: engineers as AI reviewers
24:13 Big AI spend, no P&L impact: what a CIO should do
27:03 Who owns AI transformation if not the CIO?
29:21 Productivity, cost or revenue? What AI projects actually target
31:29 One foot in the old world, one foot in the new
