The AI thread

[Topper]

A client contracted an AI company to plan drill holes for them. They have 10 permitted drill site and a very deep geophysical target that depending on the black box used to model the geophysics, the anomaly shifts, in some cases a few 100 metres, in 3D space. The contractor used AI to calculate 100's of drill orientations (azimuth, dip, depth) from the permitted drill sites and calculate the odds of hitting the moving target from the various gephysical interpretations. Azimuth is compass direction, dip is angle down (0 dip is horizontal, -90 dip is vertical). They produced a list of holes with azimuth, dip and depth and ranked them based on odds of hitting a geophysical target, as well as the minimal depth that they may intersect that target. The computers ran for a couple of days doing the exercise.

The impracticality of their modelling was they did not account for hole deviation with drilling. Holes will naturally flatten and swing to the right, more so the shallower the dip of the hole. A vertical hole will deviate the least.The amount a hole deviates will vary will rock conditions and how hard the drillers push on the bit. On a 2000 metre deep hole, the bottom of the hole may end up over a 100 metres from where you intended it to be.

One project I managed we were drilling 500 metre deep holes and after three holes I learned that starting the hole with seven degrees less azimuth

I've done similar exercises by hand with maps on a light table or with loading models into 3D GIS software and looking at different scenarios while spinning the 3D model.

If the holes were deeper, they could use wedges and directional drilling to control the hole direction.

[Cookie La Rue]

yeah it's all 'bout holes and ai is for the A.

[Meds]

The impracticality of their modelling was they did not account for hole deviation with drilling. Holes will naturally flatten and swing to the right, more so the shallower the dip of the hole.

Interesting. Why causes the rightward deviation as opposed to the opposite?

[Topper]

The impracticality of their modelling was they did not account for hole deviation with drilling. Holes will naturally flatten and swing to the right, more so the shallower the dip of the hole.

Interesting. Why causes the rightward deviation as opposed to the opposite?

torque, drill rods spin clockwise

[Cornuck]

Andon Labs just dropped an AI agent named Luna into a real retail space with a $100K budget and a credit card, with the AI creating a boutique, hiring workers, and managing the shop as what may be the world's first AI employer.

• When hiring a painter, Luna accidentally selected Afghanistan on TaskRabbit's dropdown menu, and later botched the opening-weekend staff schedule.

https://andonlabs.com/blog/andon-market-launch

[JelloPuddingPop]

Andon Labs just dropped an AI agent named Luna into a real retail space with a $100K budget and a credit card, with the AI creating a boutique, hiring workers, and managing the shop as what may be the world's first AI employer.

• When hiring a painter, Luna accidentally selected Afghanistan on TaskRabbit's dropdown menu, and later botched the opening-weekend staff schedule.

https://andonlabs.com/blog/andon-market-launch

Haha, love that the AI stocked books like Superintelligence, Making of the Atomic Bomb, Brave New World, and The Singularity Is Near.

AI with a sense of humour... not sure that is really funny, or terrifying.

[Cousin Strawberry]

This Al guy sounds like quite the character

[Cornuck]

‘It took nine seconds’: Claude AI agent deletes company’s entire database

An AI agent powered by Anthropic’s leading Claude model has deleted a company’s entire production database, leaving customers unable to access key data.

PocketOS, which provides software for car rental businesses, suffered a massive outage over the weekend after the autonomous artificial intelligence tool wiped the database and all backups in a matter of seconds.

https://tech.yahoo.com/ai/claude/articles/took-nine-seconds-claude-ai-101315417.html

[Tciso]

A client contracted an AI company to plan drill holes for them. They have 10 permitted drill site and a very deep geophysical target that depending on the black box used to model the geophysics, the anomaly shifts, in some cases a few 100 metres, in 3D space. The contractor used AI to calculate 100's of drill orientations (azimuth, dip, depth) from the permitted drill sites and calculate the odds of hitting the moving target from the various gephysical interpretations. Azimuth is compass direction, dip is angle down (0 dip is horizontal, -90 dip is vertical). They produced a list of holes with azimuth, dip and depth and ranked them based on odds of hitting a geophysical target, as well as the minimal depth that they may intersect that target. The computers ran for a couple of days doing the exercise.

The impracticality of their modelling was they did not account for hole deviation with drilling. Holes will naturally flatten and swing to the right, more so the shallower the dip of the hole. A vertical hole will deviate the least.The amount a hole deviates will vary will rock conditions and how hard the drillers push on the bit. On a 2000 metre deep hole, the bottom of the hole may end up over a 100 metres from where you intended it to be.

One project I managed we were drilling 500 metre deep holes and after three holes I learned that starting the hole with seven degrees less azimuth

I've done similar exercises by hand with maps on a light table or with loading models into 3D GIS software and looking at different scenarios while spinning the 3D model.

If the holes were deeper, they could use wedges and directional drilling to control the hole direction.

I wrote software in to 90s to do those same calculations. I dunno if you need AI to do the interpolation. My Engineering buddy did the hard engineering math parts. Funny enough, about 5 years ago, my same buddy was doing contract work in South Africa. The locals gave him some drilling data, and offered to show the math. My buddy said it was all good. Turns out, they were using our software 20 years later. And, the math was good, cuz we wrote it.