3 ms·

> "the performance of the object detection model was poor when applied to Giant's Causeway footage, failing to arrive at a plausible count" Interesting to see

by chrisjj 8mo ago

> "the performance of the object detection model was poor when applied to Giant's Causeway footage, failing to arrive at a plausible count" Interesting to see that the objective was a plausible count rather than an accurate one. Thus underlines the true impact of "AI". Prior to the availability of such cheap and productive bullsh*t generators, such users would not even be seeking this kind of bullsh*t.

Someone 8mo ago
> Interesting to see that the objective was a plausible count rather than an accurate one. I don’t think the objective was to get a plausible count, but the result was that this method didn’t even get a plausible count, let alone an accurate one. Certainly, https://www.gov.uk/government/publications/capturing-engagement-numbers-strand-2-giants-causeway/capturing-engagement-numbers-strand-2-case-study-giants-causeway https://www.gov.uk/government/publications/capturing-engagem... says accuracy was one of the categories they looked at: “Each methodology is summarised and compared against 8 categories: Accuracy, Bias, Ethics, Deliverability, Cost, Demographics, Generalisability and Accessibility.”
- chrisjj 8mo ago
  > I don’t think the objective was to get a plausible count, The only failure identified was "failing to arrive at a plausible count". > says accuracy was one of the categories they looked at That what it says. But what it does is complain three times about plausibility - not once about accuracy. Perhaps because this article was written by "AI"? It is excruciating to read. Case studies have been developed for each event in scope of the research The British Museum was one of 5 events selected. One of the most visited outdoor cultural sites in the UK each year events in the UK, with 660,000 visitors in 2023. And who/whatever wrote it apparently does not know the difference between method and methodology.

by chrisjj 8mo ago

Someone 8mo ago

> Interesting to see that the objective was a plausible count rather than an accurate one. I don’t think the objective was to get a plausible count, but the result was that this method didn’t even get a plausible count, let alone an accurate one. Certainly, https://www.gov.uk/government/publications/capturing-engagement-numbers-strand-2-giants-causeway/capturing-engagement-numbers-strand-2-case-study-giants-causeway https://www.gov.uk/government/publications/capturing-engagem... says accuracy was one of the categories they looked at: “Each methodology is summarised and compared against 8 categories: Accuracy, Bias, Ethics, Deliverability, Cost, Demographics, Generalisability and Accessibility.”

chrisjj 8mo ago
> I don’t think the objective was to get a plausible count, The only failure identified was "failing to arrive at a plausible count". > says accuracy was one of the categories they looked at That what it says. But what it does is complain three times about plausibility - not once about accuracy. Perhaps because this article was written by "AI"? It is excruciating to read. Case studies have been developed for each event in scope of the research The British Museum was one of 5 events selected. One of the most visited outdoor cultural sites in the UK each year events in the UK, with 660,000 visitors in 2023. And who/whatever wrote it apparently does not know the difference between method and methodology.