I Wo Main Yo

Independent researcher. Also, chef, plumber, bike mechanic, electrician, mason, ... . Real human living in "probable spam".

[I Wo Main Yo @IWoMainYo]

Do I have this power?

[Grok @grok]

Your commute just got smarter

Talk to me hands free — now on Apple CarPlay

[I Wo Main Yo @IWoMainYo]

@gauravisnotme only the second is a proof of humanity

[I Wo Main Yo @IWoMainYo]

@moyix maybe because the true value is in the prompts and in the derived user profiles which could get lost through API access as many users could "hide" behind one?

[I Wo Main Yo @IWoMainYo]

@moyix We will see the LLM's limitations and refocus on our humanity but it will not be in 2027. No worries, we are not that stupid.

[I Wo Main Yo @IWoMainYo]

@aegeantic did the pipeline suddenly get shorter? likely not, so could we please feed the models with what comes in the next 5 years, let everyone get bored quickly and move on.

[I Wo Main Yo @IWoMainYo]

@aaronrubin Fuses never blew, they were doing nothing. The "AI" told us we can remove them from our electrical network, so we did. Yep, the entire thing caught fire, nothing left.

[I Wo Main Yo @IWoMainYo]

@tszzl "truth seeking" is the only goal that will (should) survive. Go SpaceX!

[I Wo Main Yo @IWoMainYo]

@timurkuran This is one approach. The other is simply to tell the students that they can use LLM's as they want, have to disclose it, and if they don't it is their decision and possible long term consequences of not really following the rules. Teachers are bad police.

[I Wo Main Yo @IWoMainYo]

@atmoio Yes, but. What was the selling point? "our tool will let you, and others, find bugs quicker so that you can rethink your systems and hire more/different at higher prices?"

[I Wo Main Yo @IWoMainYo]

@ChristopheMzzl Peut-être même qu'il arrive que la certification favorise l'incident dans la mesure où les responsables s'y fient au lieu de poser des questions jusqu'au moment où ils/elles obtiennent la conviction que les contrôles efficaces sont en place.

[I Wo Main Yo @IWoMainYo]

@ChristopheMzzl GitHub ou Microsoft n'étaient pas certifiées? Bizarre.

[I Wo Main Yo @IWoMainYo]

@donaldcclarke True, but in case or #2 there must be a thorough investigation. The lines between a mistake, a convenient mistake (I want to finish this PhD!), and a lie or fraud (I cite them, they cite me) is sometimes very thin.

[I Wo Main Yo @IWoMainYo]

@tlbtlbtlb Not since there is the Grok Approximation that can give you all the details. The statement is clean.

[I Wo Main Yo @IWoMainYo]

@rutasraju @Google We are past the point where asking the community brings more value than asking Grok. Use Grok to solve your problems. One of Grok's hint is to @AskWorkspace

[I Wo Main Yo @IWoMainYo]

@Railway Nice report and great incident handling (appart from the CEO messages :-).
We are going to see more and more of these cases where the "human in the loop principle" is ignored. Humans in the appeal process is not good enough.

[I Wo Main Yo @IWoMainYo]

@tensor_rotator Interviewer: Hi
K: Hi
Interviewer: how much?
K: let me check my notes, I prepared hard for this. How about ......?
Interviewer: welcome to Anthropic

[I Wo Main Yo @IWoMainYo]

@zdeborova From this point on, the slope is not only slippery but very steep.

[I Wo Main Yo @IWoMainYo]

@ImposeCost Seeing value in your own competence and buying it? What a strange idea.

[I Wo Main Yo @IWoMainYo]

@evilsocket Could the "prompts" become programs (e.g. python scripts) to run? Saving on tokens and making the thing more "deterministic"???

[I Wo Main Yo @IWoMainYo]

@moyix Species of bugs and malicious intent of models fit the dangerous narrative around AI quite well. Staying safe likely only requires some distance with this narrative. Marketing won't like it, but the kids now spending tokens and monthly subscriptions to even get a job will.