Ip regulations for ml discoveries

<p><a class="wikilink" href="/sdl_(self-driving_labs)/">SDL (self-driving labs)</a> are tools able to make discoveries of new materials and processes that are not explicitly defined. Thanks to developments in <a class="wikilink" href="/machine_learning/">machine learning</a> and <a class="wikilink" href="/artificial_intelligence/">artificial intelligence</a>, there's a relatively high likelihood that a system can "discover" something useful. But that would trigger concerns regarding the intellectual property ownership. (See: <a class="wikilink" href="/202504281141_roadblocks_in_the_future_of_sdls/">202504281141 Roadblocks in the future of SDLs</a>). </p> <p>In the development of <a class="wikilink" href="/intellectual_property/">intellectual property</a>, there's normally a creative step. There was this discussion regarding the photo taken by a monkey: was the copyright in the hands of the animal or of the photographer. Is the triggering of the shutter "creative enough?" compared to setting up the parameters? </p> <p>And the same discussion will happen with <a class="wikilink" href="/sdl_(self-driving_labs)/">SDL (self-driving labs)</a>. Who will own the IP of a discovery made by an algorithm? The owner of the lab, the "user" who prepared it and triggered it? So far, there's little certainty about how <a class="wikilink" href="/patents/">patents</a> will interact with algorithmic discoveries.</p>