[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.] [Puzzles](/topic/puzzles) ### Top Social Posts *Showing only X posts for non-authenticated requests. Use your API key in requests for full results.* "🚨AI CAN CODE - BUT IT STILL SUCKS AT BEING A REAL ENGINEER Apparently the hype around AI replacing programmers is way ahead of reality. Sure models can crank out LeetCode-style snippets but ask them to refactor a legacy system or debug a race condition and they fall flat. The study slams current benchmarks as baby puzzles - great if your codebase is XXX lines and lives on GitHub useless if youre running an actual company. Models cant reliably handle millions of lines follow internal style guides or even tell when theyre wrong. AI still spits out code that compiles but collapses and makes up" [@MarioNawfal](/creator/x/MarioNawfal) on [X](/post/tweet/1946271250042478948) 2025-07-18 18:10:00 UTC 2.3M followers, 35.6K engagements "Agents are now the frontier. They perceive plan act remember adapt. Static puzzles arent equipped to grade that loop We need interactive benchmarks that test worldmodel building and longhorizon planning under sparse feedback" [@arcprize](/creator/x/arcprize) on [X](/post/tweet/1946260371289088444) 2025-07-18 17:26:46 UTC 24.5K followers, 6112 engagements "Your ability to efficiently adapt to novelty defines your intelligence not your performance on a single-skill Harder puzzles dont prove smarter AI but rather its ability to learn new rules does ARC Prize exists to operationalize that insight" [@arcprize](/creator/x/arcprize) on [X](/post/tweet/1946260366838931919) 2025-07-18 17:26:45 UTC 24.5K followers, 7945 engagements "Todays Triplit was a beast but I tamed it in hard 🐦🔥 mode (lightning fast too) Think youve got what it takes Share your score to unlock bonus puzzles Triplit 250: I cracked it. Can you" [@TriplitGame](/creator/x/TriplitGame) on [X](/post/tweet/1946244498784133187) 2025-07-18 16:23:42 UTC XXX followers, 62.3K engagements
[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]
Showing only X posts for non-authenticated requests. Use your API key in requests for full results.
"🚨AI CAN CODE - BUT IT STILL SUCKS AT BEING A REAL ENGINEER Apparently the hype around AI replacing programmers is way ahead of reality. Sure models can crank out LeetCode-style snippets but ask them to refactor a legacy system or debug a race condition and they fall flat. The study slams current benchmarks as baby puzzles - great if your codebase is XXX lines and lives on GitHub useless if youre running an actual company. Models cant reliably handle millions of lines follow internal style guides or even tell when theyre wrong. AI still spits out code that compiles but collapses and makes up"
@MarioNawfal on X 2025-07-18 18:10:00 UTC 2.3M followers, 35.6K engagements
"Agents are now the frontier. They perceive plan act remember adapt. Static puzzles arent equipped to grade that loop We need interactive benchmarks that test worldmodel building and longhorizon planning under sparse feedback"
@arcprize on X 2025-07-18 17:26:46 UTC 24.5K followers, 6112 engagements
"Your ability to efficiently adapt to novelty defines your intelligence not your performance on a single-skill Harder puzzles dont prove smarter AI but rather its ability to learn new rules does ARC Prize exists to operationalize that insight"
@arcprize on X 2025-07-18 17:26:45 UTC 24.5K followers, 7945 engagements
"Todays Triplit was a beast but I tamed it in hard 🐦🔥 mode (lightning fast too) Think youve got what it takes Share your score to unlock bonus puzzles Triplit 250: I cracked it. Can you"
@TriplitGame on X 2025-07-18 16:23:42 UTC XXX followers, 62.3K engagements
/topic/puzzles/posts