One-Sentence Takeaway

AI is most useful when credit union lending leaders treat it as an assistant that improves communication, analysis, and execution while people keep responsibility for judgment, decisions, and imagination.

Summary

Dan opened by showing the current state of AI-generated content and pointing out a new threshold: work that does not use AI may soon be the work that feels slightly off. The main point was practical, not futuristic. AI is already changing how people communicate, execute processes, find problems, and manage the growing amount of information in their work.

He connected that shift to his first major technology lesson: watching AutoCAD turn a three-day parking lot drawing task into a three-minute copy-and-paste task. The lesson was not that technology replaces the professional. The lesson was that powerful tools become assistants that change the speed, quality, and expectations of work.

The keynote then explained why AI is useful and why it still needs human review. AI predicts likely answers. It can help people become stronger in adjacent skills, but it can also give low-confidence answers that look confident. That is why leaders need to ask about confidence, sources, assumptions, and places where the AI made choices they may want to challenge.

For Origence Lending Tech Live, Dan brought the idea into credit union lending. The practical opportunity is not just to ask when vendors will add AI features. It is to start using AI now as a daily thinking partner, performance enhancer, and transformation coach while keeping human judgment central in member-impacting decisions.

Action Items

[ ] Ask an AI assistant the credit union lending prompt from the keynote, then compare the answer to your own team's current priorities.

[ ] Pick one communication-heavy workflow this week and use AI to summarize, rewrite, compare, or improve the work before sending it.

[ ] When AI gives an important answer, ask how confident it is, what assumptions it made, and which decisions along the way you might want to disagree with.

[ ] Treat AI like an intern for one recurring task: give instructions, review the output, improve the instructions, and learn the tool's capability ceiling.

[ ] Identify one area where AI can reduce time spent communicating, processing, or searching for issues so your team can spend more time solving, deciding, and imagining.

Key Ideas

AI is an assistant, not an automator.
The AutoCAD story matters because it shows the difference between a tool that changes how fast work happens and a tool that replaces the professional judgment behind the work.

The baseline for good work is moving.
As AI-generated content improves, people may start noticing work that seems slower, less polished, or less informed because AI did not help produce it.

AI works by probability, not judgment.
Large language models predict likely next words. That makes them useful for many tasks, but it also means people need to check confidence, sources, assumptions, and fit for the situation.

Wrong answers are part of the system.
Hallucinations are not a simple bug that will disappear. They are part of using probability-based tools trained on messy information, so review discipline matters.

Use AI to make yourself better.
The stronger pattern is not asking AI to do the whole job. It is asking AI to compare your work to better examples, identify gaps, and help you improve the next version.

People still own the hardest work.
AI can help with communication, process work, and issue detection. People still need to solve new problems, make decisions grounded in memories and hope, and imagine better futures.

Credit Union Lending AI Opportunities

From loan assembly to loan-ready flow:

• Auto-parse dealer packets into loan-ready fields.

• Validate stipulations with source-grounded extraction.

• Route exceptions through member-impact review.

From broad campaigns to timed member growth:

• Predict next needs from lending and account signals.

• Personalize offers with borrower timing models.

• Trigger dealer follow-up from application drop-off.

From scattered pilots to budget-disciplined scale:

• Prioritize high-ROI use cases before buying tools.

• Consolidate assistants with approved platform menus.

• Measure adoption and costs against ROI targets.

The larger point was that credit union lending leaders should use AI to rehearse these changes before they arrive as a surprise. The goal is not to blindly automate sensitive lending decisions. The goal is to improve productivity, member experience, and operational focus while preserving human review where it matters.

Talk Flow

AI-generated work is crossing a threshold

Dan began with an AI-generated video and used it to show how quickly AI-generated content is improving. The useful signal was that the old test, "this feels off, so it must be AI," is starting to reverse.

The AutoCAD copy-and-paste moment

He described learning drafting work by drawing parking lot lines by hand, then watching AutoCAD copy and paste the same work in minutes. That moment became the frame for AI: a practical technology shift that makes old workflows feel inefficient once you have seen the new way.

Assistant vs. automator

The AutoCAD lesson led to the keynote's core distinction. AI should be understood as an assistant that changes speed and leverage, not as an automator that removes responsibility from the person doing the work.

What GPT is doing

Dan used the "once upon a time" exercise to explain prediction, consensus, and the limits of AI. AI hears the statistical pattern in the room, which can be useful, but it does not have human intuition, judgment, ethics, memories, or hopes.

Confidence, mistakes, and review

The keynote showed why AI can be wrong and why low-confidence answers can still appear polished. The practical habit is to ask AI how sure it is, what it based the answer on, and where a person should review or challenge the answer.

Credit union lending shifts

Dan then applied the ideas directly to credit union lending. He asked AI for nine ways the work of lending leaders will change across loan origination, auto lending, member experience, productivity, scale, ROI discipline, and human-reviewed decisions.

Daily AI use before platform change

Before every vendor feature arrives, AI can already help with writing, data access, images, inbox processing, project planning, business strategy, and translation. The keynote encouraged attendees to use AI as a daily coaching conversation, not only as a feature embedded in enterprise systems.

Performance enhancer, not work-doing surrogate

Dan recommended asking AI to compare new work against past work and explain how to improve it. This makes the person better even if the AI is later removed from the process.

The intern metaphor

The talk challenged the "copilot" label and replaced it with the intern metaphor. Give AI instructions, review the work, improve the instructions, and learn where the tool is strong or weak.

The work pyramid

Dan described work as communication, process, issue detection, problem solving, decision-making, and imagination. AI is strongest in the lower layers, especially communication, while people remain essential above the line.

Communication, notes, songs, and translation

The keynote closed by showing how AI changes information exchange. Meeting notes, AI-generated songs, and near real-time translation were examples of how communication can become easier to capture, remember, share, and translate.

Closing Message - Managing in a world of more

The final message was that demands, systems, information, and expectations keep growing, but the week does not get more hours. The way through is to compress lower-level work with AI assistants so people can spend more time solving problems, making good decisions, and imagining better ways to serve members.