16 lines
1.4 KiB
JSON
16 lines
1.4 KiB
JSON
{
|
||
"HubID": "5607",
|
||
"Date": "10/26/2025",
|
||
"HubTags": [
|
||
"External Platform Posts",
|
||
"Future Map",
|
||
"iNDX CO+"
|
||
],
|
||
"Contacts": "",
|
||
"Companies": "",
|
||
"File": "",
|
||
"Image": "",
|
||
"Summary": "<h1 style=\"color:rgb(34, 34, 34);background-color:rgb(255, 255, 255)\"><a href=\"https://community.indx.earth/t/the-future-belongs-to-small-models-with-big-reasoning/51\" class=\"fancy-title\">The Future Belongs to Small Models with Big Reasoning </a></h1><p><br />All the reason why we need to develop good sources of data for AI platforms to utilize, because scraping the Internet like it is done now yields very low signal to noise. </p><p>Today’s giant frontier models are bloated with internet noise — great at mimicry, weak at real reasoning. Andrej Karpathy - OpenAI founding member - argues that the next wave won’t come from models that memorize everything, but from smaller “cognitive cores” that think first and fetch knowledge only when needed. Instead of trying to be the library, the model becomes the librarian: a compact reasoning engine that plans, decomposes problems, and selectively pulls in external information on demand. The future isn’t bigger models that swallow more data — it’s smaller, smarter systems that know how to go get the right data when it matters.</p><p><iframe width=\"500\" height=\"281\" src=\"//www.youtube.com/embed/lXUZvyajciY\" allowfullscreen></iframe></p>",
|
||
"Notes": ""
|
||
}
|