add wikispeedia link

index.html

@@ -45,7 +45,7 @@ <h2>Introduction</h2>
 					<p>
 						LLMs are trained on extremely large corpuses of texts, most of the time written by humans over decades. This enables them to generate human-like sentences and reply to questions in a sound way. But how closely do they adopt human ways of thought? Would an LLM play a game involving semantic relationships the same way humans do? Would they play it better?
 					</p> <p>
-						To answer these questions, we enlist an LLM as a participant in Wikispeedia, a game wherein players navigate from one Wikipedia page to another using as few clicks on Wikipedia page links as possible. We then compare the LLM’s results to human results, considering in particular speed, completion rates, strategic use of Wikipedia pages with many links, progression towards goal over time, and location of clicks on the page. Our findings contribute to a societal question more pressing in the past year than perhaps any time in human history: (when) will AI overtake humans’ ability to think creatively?
+						To answer these questions, we enlist an LLM as a participant in Wikispeedia, a game wherein players navigate from one Wikipedia page to another using as few clicks on page links as possible (click <a href="https://dlab.epfl.ch/wikispeedia/play/">here</a> to play). We then compare the LLM’s results to human results, considering in particular speed, completion rates, strategic use of Wikipedia pages with many links, progression towards goal over time, and location of clicks on the page. Our findings contribute to a societal question more pressing in the past year than perhaps any time in human history: (when) will AI overtake humans’ ability to think creatively?
 					</p>
 				</div>
 				<a href="#questions" class="button style2 down anchored">Next</a>