8 ms·
> Having scripted demos that hide error messages is not really sketchy. It could be part of a fraud, or it could not, but it's not really telling. They pricked
by PragmaticPulp 5y ago
> Having scripted demos that hide error messages is not really sketchy. It could be part of a fraud, or it could not, but it's not really telling.
They pricked investors fingers and implied that a Theranos machine would process the sample, but then they secretly ran the sample on a competitor's machine in another room:
> In demos of the company’s technology, investors would have their fingers pricked for testing alongside one of Theranos’ proprietary devices, but the actual test would be run in a lab away from prying eyes.
This would be the equivalent of a SaaS company setting up an API that just passed the calls through to a competitor's API, but then telling investors that their own platform was operational. You can run the pass-through API long enough to fake demos and raise money, but it's still fraud if you mislead investors to believe that you have the platform operational.
They weren't just hiding error messages or smoothing over rough edges. They were architecting fraudulent claims to investors.
- eli 5y agoPretty sure a non-zero number of AI/ML startups were actually just feeding data to humans, at least at the start.
- AndyNemmity 5y agocan confirm a few examples. the key is at the start though, they all made it there over some time.
- spaetzleesser 5y ago“At least at the start” is possibly ok if you have a path forward to get to your goal. Theranos had no path.
- spiffytech 5y ago> “At least at the start” is possibly ok if you have a path forward to get to your goal. Only if you're not misrepresenting your capabilities to investors or customers.
- AndyNemmity 5y agoIt's tricky. If you're saying what you will do, and show it as it should look, while you continue to develop it really working, and it's not endangering health like in the case of Theranos, I have a hard time seeing the issue. I've never been in a situation where it literally didn't work though over time, and before customer usage.
- spiffytech 5y agoThere was a story earlier this year about a startup that had a government contract to detect crime using AI, but turned out not to actually have any AI technology, instead appearing to pass tasks to humans: https://www.vice.com/en/article/pkd7pk/banjo-ai-surveillance-utah-contract https://www.vice.com/en/article/pkd7pk/banjo-ai-surveillance...
- woeirua 5y agoThis is not a problem if you're transparent about what you're doing. If you say, the AI/ML will do this, but we're still perfecting the algorithm so we have humans assist it right now. No problem.
- HWR_14 5y agoThere was just a report on HN about Expensify doing that. I don't know if they switched to AI, or if it's still just people.
- the_jeremy 5y agoThe first iPhone demo involved bringing in a portable cell tower, hardcoding signal strength to always show 5 bars, and repeatedly swapping out phones throughout the demo because using one through the whole presentation would have caused it to crash[0]. I believe Theranos was worse in many ways, but running misleading demos to drum up hype sounds pretty par for the SV course. [0]: https://www.nytimes.com/2013/10/06/magazine/and-then-steve-said-let-there-be-an-iphone.html?pagewanted=all https://www.nytimes.com/2013/10/06/magazine/and-then-steve-s..., https://www.techspot.com/news/54246-former-apple-engineer-reveals-secrets-behind-jobs-first-iphone-presentation.html https://www.techspot.com/news/54246-former-apple-engineer-re...
- Aloha 5y agoBringing in a COW is standard course when you have a large number of people in one place, even in 2007.