[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-how-evolving-ai-brains-broke-past-a-stubborn-75-ceiling":10,"sections":35},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":24,"tags":25,"sources":30,"feedback":34,"feedback_at":22,"cost_usd":34,"total_tokens":34},9597,"how-evolving-ai-brains-broke-past-a-stubborn-75-ceiling","How Evolving AI Brains Broke Past a Stubborn 75% Ceiling","Researchers show evolutionary search alone caps parity-task success at 75 percent unless activation functions evolve alongside neuromodulation.","A new study explains why evolving AI systems often plateau at mediocre performance, and what fixes it.\n\nResearchers ran more than 10,000 experiments on indirectly encoded neural substrates, the kind grown by evolutionary algorithms called CPPNs rather than hand designed. They found that letting neuromodulation alone adjust circuit behavior, without changing connections, still left a hard ceiling: standard monotonic activation functions capped success on parity tasks at 75%, no matter how much compute, network size, or population diversity was thrown at it. Gradient descent on the identical architecture solved the task completely, which means the limit was a quirk of evolutionary search, not a structural flaw in the network itself. The fix was giving evolution a second lever: letting it pick the activation function itself, pairing oscillating functions with parity tasks and monotonic ones with threshold tasks.\n\nThat combination pushed every one of 30 test runs to 100% success across five tasks at once, using a single evolved genotype that could still express different behaviors depending on context. It is a reminder that copying biology's tricks only halfway, mimicking neuromodulation but not neuron-type diversity, leaves real capability on the table. For anyone building open-ended or multi-task evolutionary systems, the takeaway is that the computational building blocks, not just the wiring, need to be searchable too.\n\nIt will not replace gradient descent for most practical AI work, but it narrows the long-standing gap between what evolution and backpropagation can each pull off.","[\"artificial-life\",\"evolutionary-algorithms\",\"neural-networks\",\"ai-research\"]","2026-10-02T04:00:00.000Z","2026-10-03T02:16:18.931Z","2026-10-03T02:16:24.063Z","published",null,[],"ai",[26,27,28,29],"artificial-life","evolutionary-algorithms","neural-networks","ai-research",[31],{"name":32,"url":33},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.00148",0,{"sections":36},[37,40,44,48,53,57,61,66,71,76,81,86,91,96],{"name":38,"slug":24,"count":39,"latest_published_at":18},"AI",5896,{"name":41,"slug":42,"count":43,"latest_published_at":18},"Security","security",837,{"name":45,"slug":46,"count":47,"latest_published_at":18},"Policy","policy",438,{"name":49,"slug":50,"count":51,"latest_published_at":52},"Deals","deals",317,"2026-10-01T22:00:00.000Z",{"name":54,"slug":55,"count":56,"latest_published_at":18},"Hardware","hardware",199,{"name":58,"slug":59,"count":60,"latest_published_at":18},"Science","science",171,{"name":62,"slug":63,"count":64,"latest_published_at":65},"Consumer Tech","consumer-tech",155,"2026-10-01T19:54:10.000Z",{"name":67,"slug":68,"count":69,"latest_published_at":70},"Dev Tools","dev-tools",96,"2026-10-01T16:57:03.000Z",{"name":72,"slug":73,"count":74,"latest_published_at":75},"Software","software",93,"2026-09-30T21:41:11.000Z",{"name":77,"slug":78,"count":79,"latest_published_at":80},"Startups","startups",90,"2026-10-01T21:55:22.000Z",{"name":82,"slug":83,"count":84,"latest_published_at":85},"Gaming","gaming",53,"2026-10-02T02:50:39.000Z",{"name":87,"slug":88,"count":89,"latest_published_at":90},"General","general",50,"2026-09-30T21:37:54.000Z",{"name":92,"slug":93,"count":94,"latest_published_at":95},"Reviews","reviews",31,"2026-09-28T14:31:34.000Z",{"name":97,"slug":98,"count":99,"latest_published_at":100},"How-To","how-to",7,"2026-10-01T09:00:00.000Z"]