šŸ¦ Twitter Post Details

Viewing enriched Twitter post

@rohanpaul_ai

šŸ’Š New study finds that clinical LLMs can ace medical exams yet still perform weakly on realistic clinical tasks and safety. models scored 84%-90% on knowledge exams but only 45%-69% on practice tasks and 40%-50% on safety assessments. The authors analyze 39 benchmarks with about 2.3 million questions across 45 languages and 172 specialties, and see knowledge-style exams largely saturated, with top models near 84%-90% accuracy. On practice-focused benchmarks such as DiagnosisArena, MedAgentBench, and HealthBench, success falls to about 45%-69%, showing that models often fail when asked to pick diagnoses, management plans, or recommendations in full cases. Looking at task types, factual lookup stays near 85%-93%, but clinical reasoning drops to 50%-60%, diagnostic accuracy to 45%-55%, and safety checks reach only 40%-50%. The authors argue that exam-style benchmarks are misleading proxies for clinical readiness and that deployment must rely on practice-based evaluation with strict human-in-the-loop oversight instead of autonomous use. --- pubmed.ncbi. nlm.nih .gov/41325597/

Media 1
Media 2
View on Twitter

šŸ“Š Media Metadata

{
  "media": [
    {
      "url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/1998029699797381166/media_0.jpg?",
      "media_url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/1998029699797381166/media_0.jpg?",
      "type": "photo",
      "filename": "media_0.jpg"
    },
    {
      "url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/1998029699797381166/media_1.jpg?",
      "media_url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/1998029699797381166/media_1.jpg?",
      "type": "photo",
      "filename": "media_1.jpg"
    }
  ],
  "processed_at": "2025-12-08T20:02:55.159573",
  "pipeline_version": "2.0"
}

šŸ”§ Raw API Response

{
  "type": "tweet",
  "id": "1998029699797381166",
  "url": "https://x.com/rohanpaul_ai/status/1998029699797381166",
  "twitterUrl": "https://twitter.com/rohanpaul_ai/status/1998029699797381166",
  "text": "šŸ’Š New study finds that clinical LLMs can ace medical exams yet still perform weakly on realistic clinical tasks and safety.\n\nmodels scored 84%-90% on knowledge exams but only 45%-69% on practice tasks and 40%-50% on safety assessments.\n\nThe authors analyze 39 benchmarks with about 2.3 million questions across 45 languages and 172 specialties, and see knowledge-style exams largely saturated, with top models near 84%-90% accuracy.\n\nOn practice-focused benchmarks such as DiagnosisArena, MedAgentBench, and HealthBench, success falls to about 45%-69%, showing that models often fail when asked to pick diagnoses, management plans, or recommendations in full cases.\n\nLooking at task types, factual lookup stays near 85%-93%, but clinical reasoning drops to 50%-60%, diagnostic accuracy to 45%-55%, and safety checks reach only 40%-50%.\n\nThe authors argue that exam-style benchmarks are misleading proxies for clinical readiness and that deployment must rely on practice-based evaluation with strict human-in-the-loop oversight instead of autonomous use.\n\n---\n\npubmed.ncbi. nlm.nih .gov/41325597/",
  "source": "Twitter for iPhone",
  "retweetCount": 3,
  "replyCount": 0,
  "likeCount": 9,
  "quoteCount": 2,
  "viewCount": 2892,
  "createdAt": "Mon Dec 08 13:59:36 +0000 2025",
  "lang": "en",
  "bookmarkCount": 5,
  "isReply": false,
  "inReplyToId": null,
  "conversationId": "1998029699797381166",
  "displayTextRange": [
    0,
    276
  ],
  "inReplyToUserId": null,
  "inReplyToUsername": null,
  "author": {
    "type": "user",
    "userName": "rohanpaul_ai",
    "url": "https://x.com/rohanpaul_ai",
    "twitterUrl": "https://twitter.com/rohanpaul_ai",
    "id": "2588345408",
    "name": "Rohan Paul",
    "isVerified": false,
    "isBlueVerified": true,
    "verifiedType": null,
    "profilePicture": "https://pbs.twimg.com/profile_images/1816185267037859840/Fd18CH0v_normal.jpg",
    "coverPicture": "https://pbs.twimg.com/profile_banners/2588345408/1729559315",
    "description": "Compiling in real-time, the race towards AGI.\n\nThe Largest Show on X for AI.\n\nšŸ—žļø Get my daily AI analysis newsletter to your email  šŸ‘‰ https://t.co/6LBxO8215l",
    "location": "Ex Inv Banking (Deutsche)",
    "followers": 113586,
    "following": 8364,
    "status": "",
    "canDm": true,
    "canMediaTag": false,
    "createdAt": "Wed Jun 25 22:38:54 +0000 2014",
    "entities": {
      "description": {
        "urls": [
          {
            "display_url": "rohan-paul.com",
            "expanded_url": "https://www.rohan-paul.com",
            "url": "https://t.co/6LBxO8215l",
            "indices": [
              134,
              157
            ]
          }
        ]
      },
      "url": {
        "urls": [
          {
            "display_url": "rohan-paul.com",
            "expanded_url": "http://www.rohan-paul.com",
            "url": "https://t.co/2NKnK0wIil",
            "indices": [
              0,
              23
            ]
          }
        ]
      }
    },
    "fastFollowersCount": 0,
    "favouritesCount": 54456,
    "hasCustomTimelines": true,
    "isTranslator": false,
    "mediaCount": 24922,
    "statusesCount": 61662,
    "withheldInCountries": [],
    "affiliatesHighlightedLabel": {},
    "possiblySensitive": false,
    "pinnedTweetIds": [
      "1965551636082032917"
    ],
    "profile_bio": {},
    "isAutomated": false,
    "automatedBy": null
  },
  "extendedEntities": {
    "media": [
      {
        "display_url": "pic.x.com/LSJg5qWUh5",
        "expanded_url": "https://x.com/rohanpaul_ai/status/1998029699797381166/photo/1",
        "id_str": "1998029190646607872",
        "indices": [
          277,
          300
        ],
        "media_key": "3_1998029190646607872",
        "media_url_https": "https://pbs.twimg.com/media/G7ps9n9aMAANZ_U.png",
        "type": "photo",
        "url": "https://t.co/LSJg5qWUh5",
        "ext_media_availability": {
          "status": "Available"
        },
        "features": {
          "large": {
            "faces": []
          },
          "medium": {
            "faces": []
          },
          "small": {
            "faces": []
          },
          "orig": {
            "faces": []
          }
        },
        "sizes": {
          "large": {
            "h": 768,
            "w": 653,
            "resize": "fit"
          },
          "medium": {
            "h": 768,
            "w": 653,
            "resize": "fit"
          },
          "small": {
            "h": 680,
            "w": 578,
            "resize": "fit"
          },
          "thumb": {
            "h": 150,
            "w": 150,
            "resize": "crop"
          }
        },
        "original_info": {
          "height": 768,
          "width": 653,
          "focus_rects": [
            {
              "x": 0,
              "y": 105,
              "w": 653,
              "h": 366
            },
            {
              "x": 0,
              "y": 0,
              "w": 653,
              "h": 653
            },
            {
              "x": 0,
              "y": 0,
              "w": 653,
              "h": 744
            },
            {
              "x": 57,
              "y": 0,
              "w": 384,
              "h": 768
            },
            {
              "x": 0,
              "y": 0,
              "w": 653,
              "h": 768
            }
          ]
        },
        "allow_download_status": {
          "allow_download": true
        },
        "media_results": {
          "result": {
            "media_key": "3_1998029190646607872"
          }
        }
      },
      {
        "display_url": "pic.x.com/LSJg5qWUh5",
        "expanded_url": "https://x.com/rohanpaul_ai/status/1998029699797381166/photo/1",
        "id_str": "1998029331910795264",
        "indices": [
          277,
          300
        ],
        "media_key": "3_1998029331910795264",
        "media_url_https": "https://pbs.twimg.com/media/G7ptF2NaoAALShr.png",
        "type": "photo",
        "url": "https://t.co/LSJg5qWUh5",
        "ext_media_availability": {
          "status": "Available"
        },
        "features": {
          "large": {
            "faces": []
          },
          "medium": {
            "faces": []
          },
          "small": {
            "faces": []
          },
          "orig": {
            "faces": []
          }
        },
        "sizes": {
          "large": {
            "h": 866,
            "w": 465,
            "resize": "fit"
          },
          "medium": {
            "h": 866,
            "w": 465,
            "resize": "fit"
          },
          "small": {
            "h": 680,
            "w": 365,
            "resize": "fit"
          },
          "thumb": {
            "h": 150,
            "w": 150,
            "resize": "crop"
          }
        },
        "original_info": {
          "height": 866,
          "width": 465,
          "focus_rects": [
            {
              "x": 0,
              "y": 0,
              "w": 465,
              "h": 260
            },
            {
              "x": 0,
              "y": 0,
              "w": 465,
              "h": 465
            },
            {
              "x": 0,
              "y": 0,
              "w": 465,
              "h": 530
            },
            {
              "x": 16,
              "y": 0,
              "w": 433,
              "h": 866
            },
            {
              "x": 0,
              "y": 0,
              "w": 465,
              "h": 866
            }
          ]
        },
        "allow_download_status": {
          "allow_download": true
        },
        "media_results": {
          "result": {
            "media_key": "3_1998029331910795264"
          }
        }
      }
    ]
  },
  "card": null,
  "place": {},
  "entities": {
    "hashtags": [],
    "symbols": [],
    "urls": [],
    "user_mentions": []
  },
  "quoted_tweet": null,
  "retweeted_tweet": null,
  "article": null
}