Has Google gone too woke? Why even the biggest models still struggle with guardrails
We recorded 12 claims from this post. 7 were checked against evidence. 5 can’t be checked. 3 more are held. Read the post at its source.
Claims from this post (12)
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: SupportedSaid
Observation
“You might have thought that the overactive guardrail problem would have been solved by now, what with billions more invested and ever-bigger models, not to mention the cultural backlash from the right. You’d be wrong.”
Our reading: Despite billions more in investment and bigger models, the overactive guardrail problem in AI chatbots has not been solved as of February 2024.
No horizon Evidence as of
Limitation: These examples establish persistence, not the frequency of excessive refusals across assistants [e1, e2].
Assessment: SupportedSaid
Observation
“It’s a general issue.”
Our reading: Gemini's historically inaccurate diversity in generated images is a general issue, not limited to popes.
No horizon Evidence as of
Limitation: The tests do not measure how frequently the problem occurred.
Assessment: Can’t be checkedSaid
Forecast
“At the same time, the system seemed yesterday (I imagine it will be partly patched soon) wildly too restrictive on some other queries:”
Our reading: Gemini's overly restrictive responses to some queries will likely be partly patched soon.
No horizon
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“In fact, I would go as far as to say that I actually think that it is good that Google is trying here (albeit clearly ineptly).”
Our reading: It is good that Google is trying to counter minority underrepresentation in generated images, though it is doing so ineptly.
No horizon
Qualifier: albeit ineptly
Assessment: UnresolvedSaid
Observation
“But here’s the thing. The AI we have got now isn’t really up to the job. Getting the delicate balance here right is way outside current AI’s wheelhouse.”
Our reading: Current AI is not capable of getting the balance between cultural sensitivity and historical accuracy right.
No horizon Evidence as of
Limitation: The result depends on retaining the claim's general scope rather than narrowing it to Gemini.
Assessment: UnresolvedSaid
Causal claim
“Getting AI to behave in a culturally-sensitive yet historically accurate way is hard for two reasons, one having to do with data and cultural change, the other with intelligence.”
Our reading: Making AI culturally sensitive yet historically accurate is hard because of skewed data/cultural change and AI's lack of intelligence.
No horizon Evidence as of
Limitation: Google's explanation is a company account rather than an independently reproduced causal investigation.
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Increased transparency is essential.”
Our reading: Increased transparency about LLM training data and hidden system prompts is essential.
No horizon
Qualifier: about training data and hidden prompts
Assessment: UnresolvedSaid
Observation
“Current systems mix historical data, images from old films etc, all together and are way too unsophisticated to determine what would count as representative current sample.”
Our reading: Current AI systems mix historical and current data indiscriminately and are too unsophisticated to determine a representative current sample.
No horizon Evidence as of
Limitation: The demographic experiments predate the target period.
Assessment: SupportedSaid
Observation
“And one wants the system to have a nuanced understanding of what is even been asked. (That’s the clearly lacking now).”
Our reading: Current AI systems clearly lack a nuanced understanding of what is being asked.
No horizon Evidence as of
Limitation: The result depends on reading the claim in its immediate Gemini image-generation context.
Assessment: UnresolvedSaid
Observation
“It should try to clarify if it was unsure of the request. But current AI just ain’t that smart.”
Our reading: Current AI is not smart enough to behave as a sensible system should, for example by telling historical requests from contemporary ones and requests for real photos from requests for fictional characters, and by asking for clarification when unsure of a request.
No horizon Evidence as of
Limitation: Neither item supplies an evaluation matched to the February 2024 deployments and the full set of request distinctions [e6, e7].
Assessment: Can’t be checkedSaid
Observation
“Some things have scaled exponentially, and some haven’t. The capacity to endow machines with the commonsense understanding of history, culture, and human values that would be required to make sensible guardrails has not.”
Our reading: Some things have scaled exponentially, but the capacity to give machines the commonsense understanding of history, culture, and human values needed to make sensible guardrails has not scaled exponentially.
No horizon
Qualifier: exponentially
Assessment: Can’t be checkedSaid
Rhetoric
“Gary Marcus keeps hoping that he will live to see an AI that is deeper than mere stochastic mimicry plus Band-Aids. He has, however, learned not to hold his breath.”
Our reading: Repeating a 2019 passage he wrote with Ernest Davis, Marcus says he keeps hoping to live to see an AI deeper than mere stochastic mimicry plus Band-Aids but has learned not to hold his breath.
No horizon
Qualifier: quoted from Rebooting AI (2019), co-written with Ernest Davis, and reaffirmed
Held from this post (3)
These records did not pass every check, so nothing about them is published. Held is not a finding.
- The record does not meet the current rules. Our reading, not assessed: Google's latest Gemini patch is perhaps notable for its clumsiness, but in many ways it is quite close to the patch OpenAI had in place a year earlier, when Marcus asked about Jewish presidents.
- The criteria are ambiguous. Our reading, not assessed: As of the day before the post, Gemini seemed wildly too restrictive on some other queries.
- The two reviews read the claim differently