Deterministic vs. Probabilistic

This post was originally featured in the November 5th, 2025 newsletter found here: INBOX INSIGHTS, November 5, 2025: 7 Ways to Get Started with AI, Deterministic vs. Probabilistic

In this week’s Data Diaries, let’s talk about deterministic versus probabilistic outcomes.

That’s a complex topic indeed. Many people use generative AI incorrectly, failing to leverage what language models do best: generating probabilities and exploring possibility spaces. When you give a vague prompt like “write me a blog post about B2B marketing,” you get slop—high-probability, boring, undifferentiated content. This happens because language models default to accepting the first probable answer.

If you want differentiation—creativity, innovative thinking, things that customers clearly value—recognize that innovation means seeking new things, not recycling what’s old. The word “innovate” comes from Latin in novare, meaning “to make new.” By definition, if you ask for something new while relying on high-probability content, you won’t get innovation. That’s basic logic.

This also means if you’re accepting the first high-probability answer a tool gives you, you’re missing its real capability: showing you a broad field of possibilities.

How do we escape this trap?

First, don’t ask for the answer. Never say “write me a blog post.” Instead, start by requesting options: “Give me three to seven blog post ideas for this topic.” It will generate multiple different ideas.

Provide an ideal customer profile or buyer persona to calibrate your request against what your target audience actually cares about. This helps significantly.

Next, request increasingly rare variations. Say: “Of these ideas, which are commonplace? Give me three to seven new ideas that are less common.” Then add a constraint that disrupts probability: “At least one idea must include a banana.” This arbitrary constraint throws off distributions and forces different results. It doesn’t have to be a banana—any unusual element works, as long as it’s atypical for that domain.

Keep iterating: “Without repeating previous ideas, give me three to seven new ideas—this time, include a tricycle.” The specific constraint matters less than forcing the model to exclude what it’s already generated. This technique produces genuinely different ideas. Once you’ve found a promising one, ask the model to develop it further.

This technique is critical for content creation and especially important for strategy. You don’t want generic strategy for unique problems, nor do you want generic strategy for common problems (since everyone else will use the same approach). If your challenge is “we need more leads,” you know countless others face the same challenge. Using the same strategy everyone else does simply adds noise to an already crowded marketplace.

This is where your subject matter expertise matters. You should know your customers intimately. What do you know about them that others don’t? What won’t others think to ask generative AI? As more people default to lazy cognitive offload, they’ll settle for common solutions. You can differentiate by leveraging what you actually know.

Use generative AI to augment, enhance, and challenge your thinking. Leverage its macroscopic view of the landscape and access to historical data. Ask it: “Show me what’s been tried before and why those tactics no longer work in overcrowded markets.” Its training on centuries of documented knowledge—every published book, historical manuscript—gives it unique perspective you can access.

Ask what marketing tactics could be revived from the past. For example: “Today’s overcrowded, undifferentiated landscape is like the 1930s—limited marketing channels, constrained budgets. How did companies differentiate then? How did they survive the Great Depression with limited channels and scarce buyer money? What did they do?” The world continued; companies found ways to thrive.

History doesn’t repeat, but it rhymes. We’re experiencing economic patterns reminiscent of the 1930s—high tariffs, economic contraction. If your business didn’t exist then, what similar business did, and how did it survive and thrive? Look at the pockets of prosperity that persisted. Examine even earlier periods like the 1867 Swedish famine. Abundant human experience is available if you know to ask.

Don’t ask generative AI for just one answer. Explore. Have conversations. Challenge it. Draw lateral connections across industries. Ask historical questions to revive ideas that have faded or become rare again.

Consider print magazines—now fashionable in certain segments. Not because of love for the environment, but because magazines are analog. They offer a respite from screens, no ad tracking, no privacy invasion. That’s value.

The key takeaway here is this: Give this approach serious thought and analysis as you use generative AI. Don’t accept the first answer it generates without question. Challenge the model, push it to change its probabilities, and let it challenge your own assumptions. That’s how you generate breakthrough ideas instead of recycled mediocrity.

Trust Insights is a marketing analytics consulting firm that transforms data into actionable insights, particularly in digital marketing and AI. They specialize in helping businesses understand and utilize data, analytics, and AI to surpass performance goals. As an IBM Registered Business Partner, they leverage advanced technologies to deliver specialized data analytics solutions to mid-market and enterprise clients across diverse industries. Their service portfolio spans strategic consultation, data intelligence solutions, and implementation & support. Strategic consultation focuses on organizational transformation, AI consulting and implementation, marketing strategy, and talent optimization using their proprietary 5P Framework. Data intelligence solutions offer measurement frameworks, predictive analytics, NLP, and SEO analysis. Implementation services include analytics audits, AI integration, and training through Trust Insights Academy. Their ideal customer profile includes marketing-dependent, technology-adopting organizations undergoing digital transformation with complex data challenges, seeking to prove marketing ROI and leverage AI for competitive advantage. Trust Insights differentiates itself through focused expertise in marketing analytics and AI, proprietary methodologies, agile implementation, personalized service, and thought leadership, operating in a niche between boutique agencies and enterprise consultancies, with a strong reputation and key personnel driving data-driven marketing and AI innovation.