Create accountLogin

Recent searches

No recent searches

Search options

Only available when logged in.
shakedown.social is one of the many independent Mastodon servers you can use to participate in the fediverse.
A community for live music fans with roots in the jam scene. Shakedown Social is run by a team of volunteers (led by @clifff and @sethadam1) and funded by donations.

Administered by:

Server stats:

258
active users

shakedown.social: AboutStatusProfiles directoryPrivacy policy

Mastodon: AboutGet the appKeyboard shortcutsView source codev4.3.7

#aialignment

0 posts0 participants0 posts today
IT News<p>Researchers concerned to find AI models hiding their true “reasoning” processes - Remember when teachers demanded that you "show your work" in school? Some ... - <a href="https://arstechnica.com/ai/2025/04/researchers-concerned-to-find-ai-models-hiding-their-true-reasoning-processes/" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">arstechnica.com/ai/2025/04/res</span><span class="invisible">earchers-concerned-to-find-ai-models-hiding-their-true-reasoning-processes/</span></a> <a href="https://schleuss.online/tags/largelanguagemodels" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>largelanguagemodels</span></a> <a href="https://schleuss.online/tags/simulatedreasoning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>simulatedreasoning</span></a> <a href="https://schleuss.online/tags/machinelearning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>machinelearning</span></a> <a href="https://schleuss.online/tags/aialignment" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>aialignment</span></a> <a href="https://schleuss.online/tags/airesearch" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>airesearch</span></a> <a href="https://schleuss.online/tags/anthropic" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>anthropic</span></a> <a href="https://schleuss.online/tags/aisafety" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>aisafety</span></a> <a href="https://schleuss.online/tags/srmodels" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>srmodels</span></a> <a href="https://schleuss.online/tags/chatgpt" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>chatgpt</span></a> <a href="https://schleuss.online/tags/biz" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>biz</span></a>⁢ <a href="https://schleuss.online/tags/claude" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>claude</span></a> <a href="https://schleuss.online/tags/ai" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ai</span></a></p>
IT News<p>Researchers astonished by tool’s apparent success at revealing AI’s hidden motives - In a new paper published Thursday titled "Auditing language models for hid... - <a href="https://arstechnica.com/ai/2025/03/researchers-astonished-by-tools-apparent-success-at-revealing-ais-hidden-motives/" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">arstechnica.com/ai/2025/03/res</span><span class="invisible">earchers-astonished-by-tools-apparent-success-at-revealing-ais-hidden-motives/</span></a> <a href="https://schleuss.online/tags/largelanguagemodels" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>largelanguagemodels</span></a> <a href="https://schleuss.online/tags/alignmentresearch" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>alignmentresearch</span></a> <a href="https://schleuss.online/tags/machinelearning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>machinelearning</span></a> <a href="https://schleuss.online/tags/claude3" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>claude3</span></a>.5haiku <a href="https://schleuss.online/tags/aialignment" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>aialignment</span></a> <a href="https://schleuss.online/tags/aideception" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>aideception</span></a> <a href="https://schleuss.online/tags/airesearch" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>airesearch</span></a> <a href="https://schleuss.online/tags/anthropic" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>anthropic</span></a> <a href="https://schleuss.online/tags/chatgpt" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>chatgpt</span></a> <a href="https://schleuss.online/tags/chatgtp" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>chatgtp</span></a> <a href="https://schleuss.online/tags/biz" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>biz</span></a>⁢ <a href="https://schleuss.online/tags/claude" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>claude</span></a> <a href="https://schleuss.online/tags/ai" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ai</span></a></p>
Mark Abraham<p>“We need to do empirical experiments on how these things try to escape control,” Hinton told @andersen. “After they’ve taken over, it’s too late to do the experiments.” @TheAtlantic @OpenAI <a href="https://mastodon.world/tags/aialignment" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>aialignment</span></a> <a href="https://mastodon.world/tags/ai" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ai</span></a></p>
Hobson Lane<p><span class="h-card"><a href="https://mstdn.social/@rysiek" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>rysiek</span></a></span> <span class="h-card"><a href="https://pleroma.pch.net/users/woody" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>woody</span></a></span> The first step in controlling or regulating AI is predicting what it will do next. <br>( <a href="https://mstdn.social/tags/AIControlProblem" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AIControlProblem</span></a> <a href="https://mstdn.social/tags/AISafety" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AISafety</span></a> <a href="https://mstdn.social/tags/AIAlignment" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AIAlignment</span></a> - <a href="https://en.m.wikipedia.org/wiki/AI_alignment" rel="nofollow noopener noreferrer" target="_blank"><span class="invisible">https://</span><span class="ellipsis">en.m.wikipedia.org/wiki/AI_ali</span><span class="invisible">gnment</span></a> )</p><p>And to predict what a system will do next you have to first get good at explaining why it did what it did the last time.</p><p>The smartest researchers think we're decades away from being able to explain deep neural networks. So LLMs &amp; self driving cars keep doing bad things.</p><p><a href="https://mstdn.social/tags/AIExplainability" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AIExplainability</span></a> - <a href="https://en.wikipedia.org/wiki/Explainable_artificial_intelligence" rel="nofollow noopener noreferrer" target="_blank"><span class="invisible">https://</span><span class="ellipsis">en.wikipedia.org/wiki/Explaina</span><span class="invisible">ble_artificial_intelligence</span></a></p>
Roban Hultman Kramer<p>Anyway, I keep meaning to write up a blog post on “falsehoods I have believed about measuring model performance” touching on <a href="https://sigmoid.social/tags/AppliedML" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AppliedML</span></a> issues related to <a href="https://sigmoid.social/tags/modelEvaluation" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>modelEvaluation</span></a>, <a href="https://sigmoid.social/tags/metrics" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>metrics</span></a>, <a href="https://sigmoid.social/tags/monitoring" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>monitoring</span></a>, <a href="https://sigmoid.social/tags/observability" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>observability</span></a>, and <a href="https://sigmoid.social/tags/experiments" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>experiments</span></a> (<a href="https://sigmoid.social/tags/RCTs" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>RCTs</span></a>). The cool kids would call this <a href="https://sigmoid.social/tags/AIAlignment" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AIAlignment</span></a> in their VC pitch decks, but even us <a href="https://sigmoid.social/tags/NormCore" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NormCore</span></a> ML engineers have to wrestle with how to measure and optimize the real-world impact of our models.</p>
ExploreLive feeds
About