this post was submitted on 17 Aug 2024
64 points (100.0% liked)

the_dunk_tank

15909 readers
19 users here now

It's the dunk tank.

This is where you come to post big-brained hot takes by chuds, libs, or even fellow leftists, and tear them to itty-bitty pieces with precision dunkstrikes.

Rule 1: All posts must include links to the subject matter, and no identifying information should be redacted.

Rule 2: If your source is a reactionary website, please use archive.is instead of linking directly.

Rule 3: No sectarianism.

Rule 4: TERF/SWERFs Not Welcome

Rule 5: No ableism of any kind (that includes stuff like libt*rd)

Rule 6: Do not post fellow hexbears.

Rule 7: Do not individually target other instances' admins or moderators.

Rule 8: The subject of a post cannot be low hanging fruit, that is comments/posts made by a private person that have low amount of upvotes/likes/views. Comments/Posts made on other instances that are accessible from hexbear are an exception to this. Posts that do not meet this requirement can be posted to !shitreactionariessay@lemmygrad.ml

Rule 9: if you post ironic rage bait im going to make a personal visit to your house to make sure you never make this mistake again

founded 4 years ago
MODERATORS
 

The kicker:

The system isn’t purely run on AI. It also relies on human input to help it learn and become more accurate.

And then

And so far it’s getting high marks from employees

(no employees interviewed or cited)

Who call it “Bo-Linda”

Love to regurgitate company PR material as fact

you are viewing a single comment's thread
view the rest of the comments
[–] lurkerlady@hexbear.net 10 points 2 months ago* (last edited 2 months ago)

lmao the company running this, Hi Auto, is also an israeli firm

from what i can tell they hook into google voice's api and then hook the output into an llm to respond. the order is put into their system through RAG. this is like, script kiddie shit but they have a company of 50 people doing it (with 12 actual programmers). they also advertise 90% accuracy of the order which is fucking abysmal, any legit data scientist would be ashamed of such a low rate. it should be noted that accuracy is usually measured per token, so imagine it getting 10% of your syllables wrong in your order, which can end up with an order that is very wrong and could trigger the wrong RAG (ordering something you dont want). some of this is likely due to issues with audio quality from shit ass drive thru mics, loud motors, and bad noise cancelling. you can get a much higher accuracy spooling up a random local speech+llm model at home while picking your nose