Discussion
Loading...

#Tag

Log in
  • About
  • Code of conduct
  • Privacy
  • About Bonfire
George Liquor, American
George Liquor, American
@liquor_american@universeodon.com  ·  activity timestamp 2 days ago

@reflex @davidgerard It's enough to force one to re-swallow lunch

David Fleetwood - RG Admin
David Fleetwood - RG Admin
@reflex@retrogaming.social replied  ·  activity timestamp 2 days ago

@liquor_american @davidgerard It's going extremely well in the Sharkey discord. I love how after all of that he flat out admits he's not a dev but somehow will recruit devs to review it. You know, for the project that he said he couldn't find devs to help with.

#panos #catodon #ai

Discord chat reading -

Panos: I have been completely honest and straightforward about what I'm doing, and do not plan to release a stable version if it hasn't been evaluated by actual devs. Even then, I will continue to explain any risks, and users can make up their minds for what they want to do, I do not plan to force anyone to use it. And I made an announcement exactly because I understand that I need to find people to test and evaluate it. Sorry if that disgusts you.

Hazelnoot: I'll withhold my opinions on the use of AI, but if you don't mind one unsolicited suggestion - please let an actual developer audit the code before running it in production. Federation code has extremely subtle security concerns, and the relative lack of ActivityPub code within AI training datasets means that AI coding tools are more likely to make mistakes. Of course it's your own project so you can totally choose to ignore me if you want, but this is a genuine suggestion coming from a desire to not see you / your users get hacked.

Panos: No, I'm not a developer, and you are right - but as I already said in the post, I am 100% aware of the need to have a developer audit the code, and that's the plan. BTW I already have a first interesting incident 😁  GPT says it found a bug in the stable Sharkey code - I'll paste its findings and the fix it suggests in the support channel, so you can see if it's right or it's hallucinating
Discord chat reading - Panos: I have been completely honest and straightforward about what I'm doing, and do not plan to release a stable version if it hasn't been evaluated by actual devs. Even then, I will continue to explain any risks, and users can make up their minds for what they want to do, I do not plan to force anyone to use it. And I made an announcement exactly because I understand that I need to find people to test and evaluate it. Sorry if that disgusts you. Hazelnoot: I'll withhold my opinions on the use of AI, but if you don't mind one unsolicited suggestion - please let an actual developer audit the code before running it in production. Federation code has extremely subtle security concerns, and the relative lack of ActivityPub code within AI training datasets means that AI coding tools are more likely to make mistakes. Of course it's your own project so you can totally choose to ignore me if you want, but this is a genuine suggestion coming from a desire to not see you / your users get hacked. Panos: No, I'm not a developer, and you are right - but as I already said in the post, I am 100% aware of the need to have a developer audit the code, and that's the plan. BTW I already have a first interesting incident 😁 GPT says it found a bug in the stable Sharkey code - I'll paste its findings and the fix it suggests in the support channel, so you can see if it's right or it's hallucinating
Discord chat reading - Panos: I have been completely honest and straightforward about what I'm doing, and do not plan to release a stable version if it hasn't been evaluated by actual devs. Even then, I will continue to explain any risks, and users can make up their minds for what they want to do, I do not plan to force anyone to use it. And I made an announcement exactly because I understand that I need to find people to test and evaluate it. Sorry if that disgusts you. Hazelnoot: I'll withhold my opinions on the use of AI, but if you don't mind one unsolicited suggestion - please let an actual developer audit the code before running it in production. Federation code has extremely subtle security concerns, and the relative lack of ActivityPub code within AI training datasets means that AI coding tools are more likely to make mistakes. Of course it's your own project so you can totally choose to ignore me if you want, but this is a genuine suggestion coming from a desire to not see you / your users get hacked. Panos: No, I'm not a developer, and you are right - but as I already said in the post, I am 100% aware of the need to have a developer audit the code, and that's the plan. BTW I already have a first interesting incident 😁 GPT says it found a bug in the stable Sharkey code - I'll paste its findings and the fix it suggests in the support channel, so you can see if it's right or it's hallucinating
  • Copy link
  • Flag this comment
  • Block

BT Free Social

BT Free is a non-profit organization founded by @ozoned@btfree.social . It's goal is for digital privacy rights, advocacy and consulting. This goal will be attained by hosting open platforms to allow others to seamlessly join the Fediverse on moderated instances or by helping others join the Fediverse.

BT Free Social: About · Code of conduct · Privacy ·
Bonfire social · 1.0.1 no JS en
Automatic federation enabled
Log in
  • Explore
  • About
  • Code of Conduct