Embed Notice

HTML Code

<blockquote style="position: relative; padding-left: 55px;"><section><a href="https://mastodon.social/users/FeralRobots/statuses/112048505981413819">FeralRobots (feralrobots@mastodon.social)'s status on Wednesday, 06-Mar-2024 23:19:50 JST</a><a href="https://mastodon.social/@FeralRobots" title="feralrobots@mastodon.social"><img src="https://gnusocial.jp/avatar/28525-48-20221116143827.webp" width="48" height="48" alt="FeralRobots" style="position: absolute; left: 0; top: 0;">FeralRobots</a></section><article><p>What's going on is that Anthropic "prompt engineers" have redefined self-awareness to mean 'has contextual information.' That the system is using language then allows them to delude themselves into universalizing their definition.</p><p>Saw a similar problem in AI research in the 80s: researchers might define a "frame" holding contextual info, &amp; when their program produced solutions that referenced the frame, construed that as a form of self-awareness. <br><a href="https://mastodon.social/tags/AIHype" rel="tag">#AIHype</a> <a href="https://mastodon.social/tags/Claude" rel="tag">#Claude</a> </p><p><a href="https://arstechnica.com/information-technology/2024/03/claude-3-seems-to-detect-when-it-is-being-tested-sparking-ai-buzz-online/" rel="nofollow noreferrer">https://arstechnica.com/information-technology/2024/03/claude-3-seems-to-detect-when-it-is-being-tested-sparking-ai-buzz-online/</a></p></article><footer><a rel="bookmark" href="https://gnusocial.jp/conversation/2796050#notice-5543551">In conversation</a><time datetime="2024-03-06T23:19:50+09:00" title="Wednesday, 06-Mar-2024 23:19:50 JST">about a year ago</time> <span>from <span><a href="https://mastodon.social/@FeralRobots/112048505981413819" rel="external" title="Sent from mastodon.social via ActivityPub">mastodon.social</a></span></span><a href="https://mastodon.social/@FeralRobots/112048505981413819">permalink</a><h4>Attachments</h4><ol><li><article><header><div>Domain not in remote thumbnail source whitelist: cdn.arstechnica.net</div><h5><a href="https://arstechnica.com/information-technology/2024/03/claude-3-seems-to-detect-when-it-is-being-tested-sparking-ai-buzz-online/">Anthropic’s Claude 3 causes stir by seeming to realize when it was being tested</a></h5><div> from <span>@benjedwards</span></div></header><div>Claude: "This pizza topping 'fact' may have been inserted as a joke or to test if I was paying attention."</div><footer></footer></article></li></ol></footer></blockquote>

Corresponding Notice

Embed this notice
FeralRobots (feralrobots@mastodon.social)'s status on Wednesday, 06-Mar-2024 23:19:50 JST FeralRobots
What's going on is that Anthropic "prompt engineers" have redefined self-awareness to mean 'has contextual information.' That the system is using language then allows them to delude themselves into universalizing their definition.
Saw a similar problem in AI research in the 80s: researchers might define a "frame" holding contextual info, & when their program produced solutions that referenced the frame, construed that as a form of self-awareness.
#AIHype #Claude
https://arstechnica.com/information-technology/2024/03/claude-3-seems-to-detect-when-it-is-being-tested-sparking-ai-buzz-online/
In conversationabout a year ago from mastodon.socialpermalink
Attachments
1. Domain not in remote thumbnail source whitelist: cdn.arstechnica.net
  Anthropic’s Claude 3 causes stir by seeming to realize when it was being tested
  from @benjedwards
  Claude: "This pizza topping 'fact' may have been inserted as a joke or to test if I was paying attention."

Public

Embed Notice

HTML Code

Corresponding Notice