Is AI already killing people by accident?
We recorded 14 claims from this post. 1 was checked against evidence. 13 can’t be checked. 5 more are held. Read the post at its source.
Claims from this post (14)
In the order they appear in the post. Each quotation is the speaker’s words in this post. A label such as “Supported” or “Can’t be checked” is the claim’s current assessment; its page has the evidence and the reasons.
Assessment: Can’t be checkedSaid
Forecast
“We may get little detail about AI’s role or non-role in incidents in the future.”
Our reading: The public may get little detail about AI's role in future military incidents.
No horizon
Assessment: SupportedSaid
Observation
“And of course Generative AI still has problems with common sense, as an endless of stream of examples have shown, this one involving circulating on X today:”
Our reading: Generative AI still has problems with common sense.
No horizon Evidence as of
Limitation: The main systematic test comes from a commercial vendor.
Assessment: Can’t be checkedSaid
Conditional forecast
“Meanwhile, unless and until the military does actual empirical studies on collateral damage, we won’t really know whether AI is helping or hurting.”
Our reading: Without empirical military studies on collateral damage, we won't know whether AI is helping or hurting in targeting.
No horizon Evidence as of
Only if: unless the military does empirical studies on collateral damage
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“(More broadly, the military use of AI needs to be a granular question.”
Our reading: Military use of AI should be evaluated granularly, task by task.
No horizon
Assessment: Can’t be checkedSaid
Causal claim
“Mileage may vary according to the task, and may be worse in unfamiliar situations, given the inherent tendencies of generative AI.)”
Our reading: Because of its inherent tendencies, generative AI's military performance may vary by task and may be worse in unfamiliar situations.
No horizon
Qualifier: military tasks
Assessment: Can’t be checkedSaid
Forecast
“Some may lead to further escalation (a mass killing of school children could well do that); in the worst case, a series of escalations triggered by AI-triggered mistakes could lead to a nuclear war.”
Our reading: AI-triggered military mistakes may lead to escalation and, in the worst case, could lead to nuclear war.
No horizon
Qualifier: worst case
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“In my own view the biggest problem with the algorithms targeting Gaza was not necessarily the algorithms per se (about which not much may be public) but the decision to tolerate a large number of civilian casualties as part of the targeting.”
Our reading: In Marcus's own view, the biggest problem with the algorithms targeting Gaza was not necessarily the algorithms themselves, about which not much may be public, but the decision to tolerate a large number of civilian casualties as part of the targeting.
No horizon
Qualifier: in my own view
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“And it should be incumbent on them to understand the limitations and inaccuracies of the algorithms they choose.”
Our reading: Those who use military AI should be obliged to understand its limitations and inaccuracies.
No horizon
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“We should absolutely want any AI that is used for war to be as precise and reliable as possible, minimizing casualties, but we should also never forget that those who use them are responsible for decisions about how many casualties are acceptable.”
Our reading: We should absolutely want any AI used for war to be as precise and reliable as possible, minimizing casualties, and we should never forget that those who use it are responsible for decisions about how many casualties are acceptable.
No horizon
Scope: any AI that is used for war
Assessment: Can’t be checkedSaid
Observation
“Whether or not current AI algorithms are precise (probably they aren’t), and whether not humans are involved in the specific selection of targets, those who use military algorithms bear responsibility for the outcomes they produce.”
Our reading: Current AI algorithms are probably not precise.
No horizon
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Whether or not current AI algorithms are precise (probably they aren’t), and whether not humans are involved in the specific selection of targets, those who use military algorithms bear responsibility for the outcomes they produce.”
Our reading: Whether or not current AI algorithms are precise (probably they aren't), and whether or not humans are involved in the specific selection of targets, those who use military algorithms bear responsibility for the outcomes they produce.
No horizon
Qualifier: whether or not current AI algorithms are precise (probably they aren't)
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“The race to shove AI into everything is grossly premature, because the tech fundamentally lack reliability.”
Our reading: The race to shove AI into everything is grossly premature because the technology fundamentally lacks reliability.
No horizon
Assessment: Can’t be checkedSaid
Forecast
“Meanwhile, the chance that we will get straight answers is probably close to zero.”
Our reading: The public will very probably not get straight answers about AI's role in military incidents.
No horizon
Assessment: Can’t be checkedSaid
Observation
“Altman, for his part, doesn’t seem to care, having signed a contract entirely full of holes, despite making noises about red lines.”
Our reading: Altman signed a military contract full of holes despite talking about red lines, and seems not to care.
No horizon
Held from this post (5)
These records did not pass every check, so nothing about them is published. Held is not a finding.
- The two reviews read the claim differently
- The two reviews read the claim differently
- The two reviews read the claim differently
- The two reviews reached different assessments. Our reading, not assessed: Generative AI still has serious problems with reasoning and visual cognition.
- The two reviews read the claim differently