Synthetic Data from Synthetic People: What If One Asks about a Topic Not in the Training Data?
May 13, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
At lunch yesterday with another dinobaby, one of the topics we kicked around was, “Why are businesses struggling to use AI in a way that does not cause problems?” The individual with whom I dined has more grandchildren than I. He was concerned that the business processes were speeding up. His grandchildren accept mobile devices, video content, and the velocity of their lives as “normal.”
“What does this mean for them? What does this mean for business going forward?” he asked.
I said, “Speed is an issue.” He looked at me glumly. I looked at him glumly.
I thought about this quite successful individual’s dual concerns: Business and grandchildren when I read “Market Research Is Too Slow for the AI Era, So Brox Built 60,000 Identical Digital Twins of Real people You Can Survey Instantly, Repeatedly.” The write up states:
Brox, a predictive human intelligence startup, recently announced a strategic funding round following a year where they reported 10X revenue growth. Their proposition is as ambitious as it is technical: the creation of a “parallel universe” populated by 60,000 digital twins of real, living human beings and their entire demographic profiles and consumer preferences, allowing enterprises to run unlimited experiments in hours rather than months.
The article adds:
The company, currently a lean 14-person operation, is positioning itself as the antithesis of the “insane” research industry. By replacing statistical models with behavioral replicas, Brox aims to transform how the world’s largest banks and pharmaceutical giants anticipate human reactions to high-stakes global and market-shifting events, or narrow, targeted product releases and personnel news, and everything in between. The kinds of surveys and specific questions that Brox asks its digital twins are completely open-ended and can be customized to fit any conceivable business customer’s use cases and goals.
Now back to the AI “struggle” some businesses face while at the same time young children just flow with the artefacts available to them, their parents, and their schools.
Businesses struggle because, by definition, they usually operate on procedures and methods discovered over time. When time is zipping along at hyper speed, an existing business is at a disadvantage. AI requires change, and implementing AI without “understanding” its nuances can lead to caution. Caution means going slow. Too bad, however, because the AI augmented express is moving faster.
Thanks, OpenAI. At least you were working when I requested an image. MidJourney and Venice.ai were not working for me.
The “gap” creates the grandfather’s anxiety about businesses and his grandchildren. Will IBM’s new AI powered data base administrator really “work” better than the flawed, expensive human variety? That’s a tough question to answer. Will my lunch partner’s grandchildren have “real jobs”? Answering these questions correctly seems to be difficult.
The story about AI instant surveys and digital twins blends the business question with the grandchildren question. For a traditional research company, like the old IP Sharp-type of outfit, this Brox story sounds a klaxon. Like it or not, Brox is likely to be just one of many New World Order research firms. Obviously the outputs will be faster than old-fashioned survey methods. My hunch is that these NWO outfits will be cheaper at first and then once the competition has been decimated and the funding sources complain, the rates will go up. But the speed is the fentanyl. Instant, repeatable surveys: How does an old-school market research firm compete?
Consider the jobs question. Let’s assume that one of my lunch mate’s grandchildren wants to be a social science/market researcher. The old-fashioned method is a case study in a text book. The better method is the AI way. It is clear that the grandchild who gets hired or who starts her own AI-powered market research firm is going to embrace the speed and repeatable approach.
Now let’s go back to the question I was asked at lunch, “Why are businesses struggling to use AI in a way that does not cause problems?” The answer is that the speed with which AI moves leaves people at the train station watching the lights of the last car fade. Those who are on the train or in the AI flow are not disoriented. In fact, the use of AI and digital twins is normal, logical, and obvious. This is bad news for organizations using AI. Something like 85 percent of businesses have AI and are using it internally. Yet only five percent, if I remember correctly, are pushing it out to the paying customers. Anxiety forces a conservative approach. Conservatism means a slower pace. Ergo. The speedy are in a better position to gut the old-fashioned outfits.
What if the digital twin approach to research is wrong? The answer is, “Rerun the research and try again.” The result, in my opinion, is more of the “good enough” result or “close enough for horseshoes.” Speed blurs some details. Therefore, one has to adapt to a world in which “good enough” is “excellent”.
Is there a fix? Nope. AI is another destabilizer of what dinobabies like me perceive as the optimal way to run a business and assist grandchildren. Managing AI is going to be a needed skill for businesses and parents. Developing that expertise is going to be a process.
What happens if a survey on a topic not in the training data is surveyed? Yeah, think good enough. I think that’s why my friend and I just looked at one another over our salads … glumly.
Stephen E Arnold, May 13, 2026
Comments
Got something to say?
-
- Subscribe to Beyond Search
Feature archive
News archive
-
