What is right? Or more specifically, what is the right thing to do when things go wrong? It is a question that we have grappled with since the dawn of civilization. We have devised elaborate systems of government and laws to help us navigate the muddy waters of right and wrong.
In the future technology will made it easier to understand, mitigate, and control the unexpected. The evolution of high speed communication between the “web of things” will allow our world to devise plans and continually reassess the optimal way to respond when things go wrong. Nonetheless, physics tells us that technology will never allow us to completely outrun the chaotic nature of the world. In a world where we will understand why things happened better then ever, the idea of blame will become increasingly difficult to assess, and may become better suited to highly powerful computers.
The following is a short story aimed at highlighting how technology could change how decisions of right and wrong will be made in the future. In a previous post I introduced the concept of Artificial Intelligence Agents (AIA) which will help us in our day to day lives. In addition to rather mundane tasks like a job hunt, interactions of our AIAs will also be important when things go wrong. They will share and store information in a way that will make dealing with the unexpected much easier. The continual stream of data collection performed by these AIAs will also make it much easier to understand the reasons why things go wrong, but unlike the world of today we will need to drill deeper and deeper into causality to find the reason that things go wrong. One day, it may even become impossible to find a fault at all…
20:28:24 – 20:29:13 December 18th, 2028
1st Digital Court of British Columbia, Canada
Quantum Computer D1078 Presiding
Simplified digest of case details and AI proceedings for public release
The following outlines the agreed details presented during the testimony of the AIAs of one Mr. Black and one Ms. White (names redacted for privacy).
Mr Black via AIA2023-AHGTOHDN (Clara):
The series of unexpected events that led to the tragic accident began at exactly 10:34:33 am on the date of December 18, 2028. Mr. Black’s GM (Google Motors) 2026 model x-1000 was travelling under the control of his AIA, Clara, at a speed of 85 km/h in the northbound direction in accordance with recommendations based on road conditions and local regulation.
A number of heat signatures were being tracked within 100 meters of the roadway in front of the vehicle. Size of heat signatures was consistent with large ruminants, which were later confirmed by visual data to be several common deer. Cross-referencing with the species importance archive showed that common deer are not endangered and under no special protection. Special notes show that deer are esthetically pleasing to people, and their presence has been listed as a desirable reminder of “natural beauty” by park planning authorities. Within the margin of safety, collision with the animals should be avoided.
It was assessed at this time that the animals were highly likely to remain in their immediate vicinity and no adjustments to speed or course of travel were warranted.
As the vehicle approached the area of the animals, one deer unexpectedly began to move at a high rate of speed towards the roadway. At this time it was assessed that collision with the animal remained highly unlikely, but that a slowing of the vehicle and an 18° turn of the front driving tires would minimize the chance of collision with the animal.
Given the current traffic conditions it was determined that this maneuver would achieve an approximately 97.87% chance of avoiding collision with the animal.
In order to maintain the chance of human injury below the legally stipulated threshold of 0.2% this would also require evasive manuvering on the part of vehicles travelling in the opposite direction. Vehicles travelling in the opposite direction were contacted successfully. A full breakdown of communications between the vehicles is provided in appendix A of this document. Optimal manuvering based on the developing probability profile was agreed upon at 10:34:38 am with the closest vehicle travelling in the opposing direction, that of one Ms. White. Diagram shown below.
A small amount of mixed rain and snow was falling on the road. In addition, a sudden gust of wind outside the 95% range of expected wind speeds for this day was blowing across the car. These factors conspired to create a sudden and unpredicted decrease in traction which was detected at 10:34:41. Given tractional changes, the possibility of collision between the vehicles of Mr. Black and Ms. White was determined to be in the range of 14.7%.
It was determined that these conditions justified extreme action. At 10:34:42 contact was again made with the vehicle of Ms. White and adjusted maneuvering was proposed to minimize likelihood of collision.
Ms. White via AIA2022-AGTYGJH (Jin):
Ms. White was travelling in her 2023 Tesla Model Tx under the control of her AIA, Jin, at 85km/h in the southbound direction in accordance with local regulation. At exactly 10:34:37, contact was made via distress frequencies with the vehicle of one Mr. Black. A deer had been detected moving towards the road, and evasive maneuvering would be required to minimize the possibility of collision between the vehicle of Ms. White and that of Mr. Black
The direction and speed of the moving deer was confirmed with sensory data feed from Ms. White’s vehicle. Maneuvering consistent with that shown in figure 1 was agreed upon at 10:34:38, and a 1.4g deceleration and 19.6° turn of the front tires was initated at this time.
At exactly 10:34:39, an updated information feed from the vehicle of Mr. Black indicated that unexpected changes in the traction conditions brought the likelihood of collision to 14.7%. Recalculation of this value showed that likelihood of collision was increasing as more sensory data was fed into the developing model of events. It was agreed that drastic action needed to be taken at this time.
Mr Black via AIA2023-AHGTOHDN (Clara):
A number of simulations of head on collisions between the vehicles of Mr. Black and Ms. White were run to examine the possible outcomes. It was determined at the time that injury to both parties was highly likely, and severe injury or death to one or both parties was unacceptably high.
To avoid the possibility of head-on impact between the vehicles, an alternative model wherein the vehicle of Ms. White would move completely off of the road surface was proposed. Under this model the likelihood of loss of control and damage to Ms. White’s vehicle was determined to be high, although the likelihood of severe injury or death to Ms. White remained below the threshold of 0.2%.
At 10:34:40, the alternative model of action was proposed to Ms. White.
Ms. White via AIA2022-AGTYGJH (Jin):
Between the time of 10:34:39 and 10:34:40, several simulations of the developing situation showed the possibility of head on collision and injury to be unacceptably high. This was confirmed by simulations run by the AIA of Mr. Black.
An alternative model wherein Ms. White would leave the road surface completely was proposed by Mr. Black. Simulations agreed upon by both parties showed a high potential that the vehicle of Ms. White would lose control and suffer damage, although the probability of severe injury or death remained below the acceptable threshold of 0.2%.
At 10:34:41, it was agreed that this alternative model of action would minimize potential for injury and property damage in the developing situation.
At this time an additional 12° movement of the front tires was initiated to guide Ms. White’s model Tx off of the road surface. An optimized adaptive breaking pattern was used to decrease the forward velocity of the car while maintaining control of the vehicle.
At 10:34:42, as Ms. White’s vehicle began to steer away from the road surface there was a highly unanticipated failure of part number 371X-442 in the front axle assembly. Calculated stresses and sensor reading for the front left wheel were well within the defined safety margins and this type of failure was of very low probability (<0.001%). It is deemed likely that a molecular fault in the part was the reason for the failure.
The failure of part 371X-442 led to a catastrophic failure of the entire wheel assembly. This in turn caused the front axle of the vehicle to dig into the soft shoulder which initiated a transverse roll of Ms. White’s vehicle. As the vehicle began to roll, safety measures within the cabin of the vehicle were deployed strategically in order to minimize injury to Ms. White.
During the high velocity rolling of her vehicle, Ms. White was subjected to high G forces that caused contusions as her unrestrained limbs came into contact with solid parts of the vehicle. Ms. White suffered an extreme fracture of her right femur, and a cracked cervical vertebrae. In addition to the blunt trauma, the extreme stress of the incident caused an undiagnosed weakness in one of Ms. White’s cerebral blood vessels to rupture, causing blood to begin to leak into Ms. White’s left frontal lobe.
At this time the microimplants in Ms. White’s body were activated to release factors targeting the burst vessel. NanoGens were activated at the specific site within 12 milliseconds of the blood vessel rupture but because Ms. White was not equipped with any specialized mechanisms to deal with cerebral hemmorhage they were of limited use. Ms. White’s NanoGens were able to place Ms. White in an unconscious state and slow blood flow to her brain in order to limit damage.
Ms. White’s vehicle finally came to stop after rolling one and a half times, coming to rest on its roof.
Mr Black via AIA2023-AHGTOHDN (Clara):
Mr. Black’s vehicle regained control following a short skid and came to a stop approximately 78 metres from where Ms. White’s vehicle had rolled over and come to a rest on its hood. Mr. Black was instructed to exit the vehicle and approach Ms. White’s vehicle.
All traffic on the road was warned of the accident and requested to come to a stop. The next closest vehicle, carrying the Green family, was requested to carefully approach the scene and provide assistance as necessary.
As Mr. Black approached the vehicle of Ms. White, he was informed of the medical situation of Ms. White. Consistent with best medical practices for a cerebral hemmorhage, Mr. Black was instructed to carefully let Ms. White down from her restraints so as to limit blood flow to her head.
Upon his arrival at the accident scene, Mr. Black found Ms. White to be unconscious. Mr. Black was then instructed to brace Ms. White as her restraints were released, and he carefully laid her down on the ground beside the accident scene.
The rapid response micro-medical team arrived within 3 minutes of the accident. A nanoswarm of medical robots was released and began treatment of Ms. White at precisely 10:38:09. A larger response team capable of transporting Ms. White to hospital was present on the scene by 10:43:22.
Ms. White via AIA2022-AGTYGJH (Jin):
Ms White was transported to the closest fully equipped medical facility where medical treatment was administered. It is expected that Ms. White will make a recovery over the next year, but she is expected to suffer permanent memory loss.
At this time, Ms. White is seeking damages for her lost income during recovery and for her pain and suffering as a result of memory loss.
Quantum Computer D1078 Ruling:
Following 10 seconds of testimony, 39 seconds of deliberation, the honourable D1078 has come to the following ruling consistent with case precedents as listed in the legal appendix:
(1) It is determined that Mr. Black via his AIA (Clara) acted appropriately and rationally throughout the incident. Probabilities generated by simulations described at various points during the accident are consistent with those calculated by more complex modelling after the accident. It is determined that Mr. Black should not be held at fault for this accident. Nonetheless, his own avoidance of injury occurred thanks to the selfless actions of Ms. White via her AIA (Jin). Thus, it is determined that Mr. Black should shoulder 18% of any financial burden placed upon Ms. White.
(2) It is determined that the Parks service should shoulder 13% of the financial burden incurred by Ms. White. Furthermore, it is ordered that the Parks service should investigate their policies regarding large animals near the roadway. In particular, it is suggested that the Parks service reassess their policies in terms of vehicle speed when near large animals.
(3) It is determined that the manufacturer of part 371X-442 should shoulder 28% of the financial burden incurred by Ms. White. The manufacturer is also ordered to launch an immediate study to assess the integrity and consistency of part 371X-442. Further action in the form of product recalls or changes in production methods will be determined following this investigation.
(4) It is determined that Ms. White’s health monitoring provider should shoulder 33% of the financial burden placed on Ms. White. According to medical testimony recorded in the appendix, the weakness in Ms. White’s cerebral blood vessel should have been easily diagnosed. The presence of hemmorhage mitigation technology in Ms. White’s implants could have significantly decreased damage incurred to Ms. White, likely bringing it below the threshold of detection. This company is ordered to undergo a complete audit of their policies to identify the flaws in their systems that allowed this accident to happen.
Summary of Judgement: It is determined that the decisions made by entities cited above in the time leading up to and during the accident were made consistent with best practices. Nonetheless, harm has been done to Ms. White and blame has been determined as outlined above. It is the sincere opinion of this court that the chance of such an accident occurring can be significantly decreased through the investigations and reforms outlined above.
great story, thanks for this. Gives rise to interesting discussions regarding the “statistical value of a life”, and the fact that we will have to begin making much more comprehensive efforts to translate our values/ethical stances into numerical representations. Also, we’re a long way off from the point where our “web of things” is developed enough to provide such exhaustive and super-fast feedback, but when it hits the scene- humanity will have some reckoning to do regarding our free-will being subsumed by technological determinism
Great point about the fact that we are going to have to really start putting a value on life. We do it now, we just don’t talk about it, eventually we will have to let go of this collective delusion though.
Fantastic and engaging story! This is my favorite post yet because of it! I loved the diagrams as well; really added to the believability and helped with comprehending the series of events. Keep up the good work 🙂
I forgot it was still 2013!