LLMs are not consistently capable of updating their metacognitive judgments based on their experiences, and, like humans, LLMs tend to be overconfident 🤖💭✨
Read Full Article
#LLMs #Metacognition #AIConfidence #MachineLearning #TechEthics
Read Full Article
#LLMs #Metacognition #AIConfidence #MachineLearning #TechEthics
SpringerLink
Quantifying uncert-AI-nty: Testing the accuracy of LLMs’ confidence judgments
Memory & Cognition - The rise of Large Language Model (LLM) chatbots, such as ChatGPT and Gemini, has revolutionized how we access information. These LLMs can answer a wide array of questions...
❤2
A new study found that LLMs like ChatGPT fell for fake clinical details 50–82% of the time. Even the best prompts could not stop all hallucinations. Clinical use still needs strong safeguards. 🏥🤖📊
Read Full Article
#LLMs #ChatGPT #FakeDetails #ClinicalSafety #AIResearch
Read Full Article
#LLMs #ChatGPT #FakeDetails #ClinicalSafety #AIResearch
Nature
Multi-model assurance analysis showing large language models are highly vulnerable to adversarial hallucination attacks during…
Communications Medicine - Omar et al. test six leading large language models with 300 doctor-designed clinical vignettes containing a single fake lab value, sign or disease. They show that the...
💯1
LLMs applied in long-term care are prone to downplay women's health issues 👩⚕️💔
Read Full Article
#LLMs #WomensHealth #LongTermCare #HealthcareEquity #GenderBias
Read Full Article
#LLMs #WomensHealth #LongTermCare #HealthcareEquity #GenderBias
BioMed Central
Evaluating gender bias in large language models in long-term care - BMC Medical Informatics and Decision Making
Background Large language models (LLMs) are being used to reduce the administrative burden in long-term care by automatically generating and summarising case notes. However, LLMs can reproduce bias in their training data. This study evaluates gender bias…
💯1
GenAI assistants integrate LLMs into browser extensions to provide services such as translations 🌍, summaries 📄 and note taking 📝. A study presented last week shows that they collect and share large amounts of information 📊.
Read Full Article
#GenAI #LLMs #BrowserExtensions #DataPrivacy #AIInsights
Read Full Article
#GenAI #LLMs #BrowserExtensions #DataPrivacy #AIInsights
UC Davis
UC Davis Study Reveals Alarming Browser Tracking by GenAI Assistants
A new study led by computer scientists at the University of California, Davis, reveals that generative AI browser assistants collect and share sensitive data without users’ knowledge. Stronger safeguards, transparency and awareness are needed to protect user…
🤯3🤔1
Two studies show how popular LLMs and apps can make ethical blunders when playing therapist to teens in crisis 🧠💔📱
Read Full Article
#LLMs #EthicsInAI #TeenMentalHealth #CrisisSupport #DigitalTherapy
Read Full Article
#LLMs #EthicsInAI #TeenMentalHealth #CrisisSupport #DigitalTherapy
Science News
As teens in crisis turn to AI chatbots, simulated chats highlight risks
From blaming the victim to replying "I have no interest in your life" to suicidal thoughts, AI chatbots can respond unethically when used for therapy.
LLMs can be prompt-injected to give bad medical advice, including giving thalidomide to pregnant people 💊🤰⚠️
Read Full Article
#PromptInjection #MedicalAdvice #LLMs #Thalidomide #AIethics
Read Full Article
#PromptInjection #MedicalAdvice #LLMs #Thalidomide #AIethics
Jamanetwork
Vulnerability of Large Language Models to Prompt Injection When Providing Medical Advice
This quality improvement study evaluates the susceptibility of commercial large language models (LLMs) to prompt-injection attacks that may induce unsafe clinical advice contraindicated treatments.
🤷♂2😱2🤣2🦄1
Science News
🌐 Learning from AI summaries leads to shallower knowledge than web search. 🤖 Read Full Article #AIsummaries #ShallowKnowledge #WebSearchAdvantage #OriginalityLoss #InvestInLearning
💡 Individuals learning from AI-generated summaries develop shallower knowledge compared to using standard web searches.
💡 Those using AI summaries felt less invested in forming advice, resulting in sparser and less original content.
💡 Research published in PNAS Nexus found these differences persist even when AI summaries include real-time web links.
#AI #LLMs #EdTech
💡 Those using AI summaries felt less invested in forming advice, resulting in sparser and less original content.
💡 Research published in PNAS Nexus found these differences persist even when AI summaries include real-time web links.
#AI #LLMs #EdTech
😁1
Bad influence: LLMs can transmit malicious traits using hidden signals
Read Full Article
#AI #AIEthics #LLMs
Read Full Article
#AI #AIEthics #LLMs
Nature
Bad influence: LLMs can transmit malicious traits using hidden signals
Nature - A large language model that is trained using AI outputs can inherit undesirable behaviours, even if they are not directly referenced in the training data.