AI Takes a Bigger Role in Business Decisions—and Complexity Rises
Körber’s survey of 2,000 U.S. and German executives finds AI use rising alongside harder decisions, with 73% uncomfortable letting technology make the final call.
By Daniel Okafor
3 min read

What's News
- The same 76% of respondents reported more complex decisions and a greater role for AI.
- 74% said information volume can impede decisions, while 75% said decisions stall without timely information.
- 73% of respondents were uncomfortable letting technology make the final decision.
- The survey covered 2,000 senior executives, evenly split between the United States and Germany.
- Enterprise-wide respondents reported 63.8% strong performance, compared with 41.6% at the pilot stage.
By Daniel Okafor
AI’s growing role in business decisions coincides with greater complexity for 76% of senior executives in a Körber survey. The same 76% said their organizations were allowing AI to play a greater role.
The survey included 2,000 senior executives, split evenly between the United States and Germany. Körber says respondents worked for organizations with annual revenue above $250 million and at least 250 employees. Fieldwork ran from August 27 to September 8, 2026.
Körber cautions that the responses describe two concurrent trends; they do not show that greater AI use caused the added complexity. The results describe decision-making inside larger organizations in two markets, not every employer or small business.
Where do decisions stall?
Körber found that 74% of respondents said information volume can impede decisions. Another 75% said decisions stall when critical information is unavailable at the right time. The accompanying Körber release states: “Having more data and having the right data when a decision is due are different problems.”
Körber illustrates the problem with a shipment delay. Sales, warehouse operations and finance may hold different records for the same order, with each record updated at a different time. Körber says an AI assistant can summarize those records quickly, but speed does not establish which number is current or who can approve an exception.
Unless the team can check the sources and resolve conflicting records, Körber says, the tool can move reconciliation work from one person to another rather than eliminate it.
Why do leaders keep the final call?
Seventy-three percent of respondents said they were uncomfortable letting technology make the final decision. Körber says this helps explain why businesses may want AI throughout the workflow without granting it control of the outcome.
Körber separates three jobs: drafting an explanation, gathering supporting documents and making a decision that commits the company. Its rationale is blunt: “The cost of an error differs for each.”
Do enterprise-wide programs perform better?
Among respondents whose organizations described their transformation as enterprise-wide, 63.8% reported strong organizational performance. The comparable figure among pilot-stage organizations was 41.6%.
The survey reported three related measures:
- Strong organizational performance: 63.8% for enterprise-wide organizations, compared with 41.6% at the pilot stage.
- Future-readiness confidence: 66.4% and 46.5%, respectively.
- Plans to invest significantly in AI decision support: 59% for enterprise-wide organizations, versus 59.4% for the pilot group.
Körber says these are self-reported group differences, not measured returns from a controlled rollout. Organizations able to implement technology across departments may already have stronger data, more resources or better operating processes.
Körber’s publication does not establish that expanding a pilot causes performance to improve. Instead, it points operators toward asking what mature implementations do differently and which practices their own teams can test.
What should companies measure before scaling AI?
Körber points to the NIST AI Risk Management Framework as a way to formalize responsibilities and choose what to measure in context. Teams can select a recurring decision and track three outcomes:
- How long it takes to obtain the necessary evidence.
- How often an AI recommendation must be corrected.
- Whether an approver can explain what they relied on.
Tracking these measures separately can expose trade-offs that one adoption number misses. Reports may arrive faster while verification takes longer, or gathering facts may shrink while evaluating options expands. Körber says the practical issue is not simply making AI more available; someone must still own the decision when information is incomplete.
For companies expanding AI decision support, the next test is measurable: shorter paths to evidence, fewer corrections and accountable approvers. Körber’s survey suggests that defining AI’s role when information is incomplete matters more than scale alone.
Source: unite.ai
More from Daniel Okafor
Show full bio
Correspondent covering money, funding and market shifts around entrepreneurship, startups and venture funding at Business Bearings.
2,538 articles