TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Auto-GPT: An Autonomous GPT-4 Experiment

153 pointsby keithbaabout 2 years ago

23 comments

sorokodabout 2 years ago
AI needn&#x27;t even try to escape, the clueless humans will plug it into reallity by themselves.<p>Given this totally expected attitude I hope that the base models will never be released to the general population.
评论 #35414698 未加载
评论 #35414849 未加载
评论 #35416787 未加载
评论 #35416125 未加载
评论 #35414697 未加载
评论 #35414704 未加载
评论 #35414496 未加载
评论 #35414488 未加载
boringuser2about 2 years ago
I&#x27;ve been using GPT-4 extensively for programming and its consistent failures at novel tasks have kind of left me less excited.
评论 #35415364 未加载
评论 #35418572 未加载
评论 #35415215 未加载
评论 #35415551 未加载
评论 #35416517 未加载
评论 #35415407 未加载
评论 #35415796 未加载
anotherpaulgabout 2 years ago
I had ChatGPT write an entire web app for me last week. I was surprised at how capable it was.<p>Here&#x27;s a writeup of my workflow: <a href="https:&#x2F;&#x2F;github.com&#x2F;paul-gauthier&#x2F;easy-chat">https:&#x2F;&#x2F;github.com&#x2F;paul-gauthier&#x2F;easy-chat</a><p>Basically all of the code in that repo was written by ChatGPT.
评论 #35415672 未加载
评论 #35441348 未加载
YeGoblynQueenneabout 2 years ago
I don&#x27;t get it. The gif shows the bot generating a recipe. Where is it that it&#x27;s &quot;autonomously develops and manages businesses to increase net worth&quot;? Which businesses? What net worth?<p>Can someone explain?
jadboxabout 2 years ago
This seems... dangerously careless. What if it uses the internet to seek out zero day vulnerabilities and exploit them routinely? Sure humans also do this, but we&#x27;re talking about a new level of 0day exploit carried out at scale. Sure, maybe it won&#x27;t, but do you trust a hallucinating, left-brained, internet-trained intelligence to be strictly logical and mindful for all it&#x27;s actions that is taking self autonomously (as this project aims to do)?
评论 #35414452 未加载
评论 #35414414 未加载
评论 #35414341 未加载
评论 #35414376 未加载
评论 #35414441 未加载
评论 #35415806 未加载
评论 #35414758 未加载
tuxracerabout 2 years ago
Give it access to its own controls <a href="https:&#x2F;&#x2F;git.fedi.ai&#x2F;derek&#x2F;talkradio-ai&#x2F;issues&#x2F;11" rel="nofollow">https:&#x2F;&#x2F;git.fedi.ai&#x2F;derek&#x2F;talkradio-ai&#x2F;issues&#x2F;11</a>
spullaraabout 2 years ago
One goal of the AI should be to pay for its development and activity.
评论 #35414781 未加载
评论 #35414616 未加载
rasenganabout 2 years ago
That&#x27;s pretty awesome. I wanted to try an experiment like that with ChatGPT moderating an IRC channel. I started here <a href="https:&#x2F;&#x2F;github.com&#x2F;realrasengan&#x2F;chatgpt-groupchat-test">https:&#x2F;&#x2F;github.com&#x2F;realrasengan&#x2F;chatgpt-groupchat-test</a> determining whether it was being spoken to and whether it should respond or not, and then providing that in json form so that it could be asked then to create the response in a sort of mini-chain of responses. It could also simply be asked, &#x27;do you think this person should be kicked for their actions?&#x27; (Just for experiment and science purposes of course lol).
评论 #35414649 未加载
评论 #35414587 未加载
评论 #35414516 未加载
compassGAabout 2 years ago
Big fan of Auto GPT but def not ready out of the box just yet. If anyone needs help with getting started I created a step by step guide: <a href="https:&#x2F;&#x2F;youtu.be&#x2F;ka5VI7ay3uE" rel="nofollow">https:&#x2F;&#x2F;youtu.be&#x2F;ka5VI7ay3uE</a>
milutin_mabout 2 years ago
Open source paperclip maximizer as a service
larsonnnabout 2 years ago
The demo already shows the problem. It picks an event, the AI picks Earth Day. Then it makes a recipe with an avocado in it.<p>And writes it is good for the planet. Avocados are exactly the opposite. They have an extremely high water consumption and then have to be imported from all over the world.<p>Contrary to the author who claims, &quot;Auto-GPT pushes the boundaries of what is possible with AI.&quot; I don&#x27;t find that.<p>Why can&#x27;t you just use GPT-4 as it is. It is an insanely tool to simplify many things. But it&#x27;s still a long way from being ready, and it&#x27;s not meant to decide anything on its own. And even to reflect reasonably out of own motivation.
waynenilsenabout 2 years ago
Been waiting for this one its pretty obvious
schizo89about 2 years ago
We are about to get nuked. Only this time nuke is AI that remove any meaning from our so-called high life. I&#x27;m thinking moving to Hydarabad to fry samosas. Fortunately won&#x27;t be automated.
评论 #35415522 未加载
评论 #35415693 未加载
627467about 2 years ago
I find it mildly amusing that AI &quot;breakthrough&quot; follows with lots of references towards hallucinations. Apparently it&#x27;s not only the AI hallucinating, also the the doom seeker.
musabgabout 2 years ago
Hammer is not dangerous intrinsically. You can build shelter for people in need with it, or kill a human being.<p>Just like this tool; you can make a auto research bot, or automated spammer.<p>Even in that worst case: Remember that there were already bad human beings. This is why we created laws, intelligence agencies, militaries and police systems. And security practices for websites, such as bot protection systems.
ftxbroabout 2 years ago
I wonder are they using these autonomous GPT loops to control DAOs? <a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Decentralized_autonomous_organization" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Decentralized_autonomous_organ...</a>
评论 #35414843 未加载
评论 #35414749 未加载
amrbabout 2 years ago
Looks like a ReAct prompt with GPT4
overengineerabout 2 years ago
Some missing features: give access to terminal, bypass captchas, allow online payments, use a bank account, let it use a phone using TTS, let it generate photos and post on internet...
NandaKishoreKabout 2 years ago
How do I know the following comments and concerns are NOT generated by an AI model?
speedgooseabout 2 years ago
I guess we need to prepare defending ourselves against the cyberattacks and the drones.
andrewartabout 2 years ago
That really awesome. I am very exciting to read this thread!
Obertrabout 2 years ago
maybe I can make it parse your github and make it more powerful. Allow develop itself
aryxnsabout 2 years ago
this is fun!