B2B Insights Podcast

The synthetic data section contains some genuinely useful, practitioner-relevant content—especially the film revenue study and the internal Superpowers dataset test—but much of the episode is a surface-level review of well-circulated AI concerns (hallucinations, IP, bias, energy) that any attentive reader of tech news would already know. There is a moderate amount of filler, repetition, and setup throat-clearing throughout.

“when you take out all the sequels, then the correlation that you see drops down to 0.15, which means it's barely better than a random guess”

“we simulated a situation where we could only get to 2/3 of fieldwork and we would get an extra third of the fieldwork from synthetic data...we found that in most situations we were just better off stopping early”

The episode is largely a synthesis of third-party reports (Deloitte, UN, Gartner, Nature) with little first-principles reasoning or contrarian argument; the interpolation-vs-extrapolation framing for synthetic data is the clearest original conceptual contribution, and the internal dataset test adds modest fresh evidence, but most takes are derivative.

“the generative AI is a really eager intern. They really want to please you. They don't want to leave you with nothing.”

“there's a strong bias towards a continuation of the status quo. So it's very unlikely to pick up on trends that were only just starting to emerge”

Thomas and Louise are mid-level practitioners at a market research firm (Senior Research Manager and Research Director), which gives them relevant applied experience, but they are not senior industry leaders, recognised AI experts, or operators who have deployed AI at meaningful scale; the episode's framing as an internal knowledge-share rather than a deep expert interview reflects this ceiling.

“one of the things I've been tasked with recently was sort of looking into the potential for usage of AI within the company and to assess, like, what we can use it for”

“we took our Superpowers data set and we basically tested it versus tested how well augmenting the data with synthetically generated additional responses worked”

The episode earns above-average marks for specificity: it cites named sources, concrete market-size figures, correlation coefficients from a named study, and an internal dataset experiment with quantified outcomes; however, some numbers are contradictory or loosely attributed, and the environmental statistics are borrowed wholesale from secondary reports without interrogation.

“The Grandview Research estimates it's about, it's at about US$164 million unFortune Businesses Insights thinks it's about US$289 million million US dollars. So pretty big numbers. And they both predict growth rates of over 30% CAGR”

“the simulated revenue had a very high correlation of 0.75 with the real world revenue...The correlation between the predicted revenue and the actual revenue drops to only 0.43...when you take out all the sequels, then the correlation that you see drops down to 0.15”

This is a co-presented lecture rather than a conversation: the hosts alternate delivering pre-prepared sections with no push-back, no follow-up questions, and no productive disagreement; transitions are purely additive affirmations that add no analytical value.

“Thanks, Thomas. Yeah, I think the topic of synthetic data is a really interesting and relevant one at the moment.”

“Thanks, Thomas. And the other concern, I think we need to keep in Mind here is a broader topic”