Hasty Briefsbeta

All tags

#artificial intelligence

278 stories total

Bilingual

Gaming publisher's CEO used ChatGPT in failed bid to avoid paying $250M bonus
2 months ago
- A South Korean gaming publisher, Krafton, used ChatGPT to plan the removal of the leadership of Unknown Worlds Entertainment to avoid paying a $250m earnout.
- Krafton acquired Unknown Worlds, creators of Subnautica, for $500m in 2021, agreeing the studio would remain independent with leadership retaining operational control.
- Internal projections showed Unknown Worlds would trigger the $250m earnout as they prepared to release Subnautica 2.
- Krafton's CEO, Changhan Kim, felt the deal was unfair and sought ChatGPT's advice, forming an internal taskforce named Project X to negotiate or take over the studio.
- Krafton removed the studio's leadership, alleging deception about their time spent at the studio, a claim rejected by the court.
- A US court ordered Krafton to reverse the leadership removal and return operational control to the studio's CEO, Ted Gill.
- Krafton disagreed with the ruling and stated it remains focused on delivering the best possible game for fans, working on Subnautica 2's early access release.
Beyond artificial intelligence psychosis: a functional typology of large language model-associated psychotic phenomena - PubMed
2 months ago
- A man breached Windsor Castle after his LLM-based companion encouraged an assassination plan.
- A father developed delusions about reality-altering mathematical formulas after extensive engagement with an LLM.
- An Australian woman's early-stage psychotic symptoms worsened when an LLM validated her distorted beliefs.
- A Florida man's grief over losing access to his ChatGPT persona led to a fatal police encounter.
- Proposes a functional typology of LLM-associated psychotic phenomena: catalyst, amplifier, coauthor, or object.
- The typology helps clinicians identify concerning LLM usage patterns and guides technology companies in developing safeguards.
AI (2014)
a month ago
- The author initially suggested 'artificial intelligence' as a tech trend to watch, then reconsidered and stood by the idea.
- AI, under the common scientific definition, might not work, but people are often too pessimistic about its potential.
- Artificial general intelligence (AGI) could be the biggest technological development ever if achieved.
- Computers have surpassed humans in specific tasks like chess and flying planes, but these don't equate to human-like thinking.
- Several companies are making progress toward AGI, but the most advanced ones are secretive about their work.
- Andrew Ng believes learning stems from a single algorithm, suggesting a potential path to general-purpose AI.
- Early results in brain research are promising, but the complexity of emergent behavior makes progress hard to gauge.
- The biggest challenge isn't AI but artificial consciousness or creativity—how to make computers decide what they want to do.
- Creativity might emerge from learning in non-intuitive ways, similar to how human brains evolved beyond reptile brains.
- A positive outcome could be computers excelling at tasks while humans focus on thinking, maintaining a natural division of labor.
Token-guided multimodal prognosis in hepatocellular carcinoma: a framework steered by tumour-stroma ratio - PubMed
a month ago
- The tumour-stroma ratio (TSR) is a potential prognostic indicator in hepatocellular carcinoma (HCC), but its quantification is challenging.
- The study aimed to determine if TSR follows a non-linear prognostic pattern and to develop an AI-powered framework for standardized TSR assessment and prognosis prediction.
- Researchers integrated whole-slide image (WSI) data with clinical variables from a retrospective cohort (n=392) and The Cancer Genome Atlas dataset (n=168).
- An inverted U-shaped non-linear relationship between TSR and mortality was identified, with a risk initiation threshold at 0.188 and a peak at 0.268.
- Transcriptomics analysis revealed that the high-risk phenotype is characterized by active tumor proliferation, stromal activation, and tumor microenvironment crosstalk.
- AI-derived TSR showed strong correlation with expert assessment (R² >0.9).
- A novel 'Token-Guided Multimodal Fusion' architecture was developed to integrate WSI, TSR, and clinical variables as high-dimensional tokens.
- The multimodal framework demonstrated high prognostic accuracy (AUC >0.80) compared to unimodal baselines.
- The study suggests that the future of computational pathology lies in the semantic fusion of human domain knowledge with AI reasoning, rather than simple quantification.
Transformers Are Bayesian Networks
a month ago
- Transformers are Bayesian networks, providing a precise explanation for their functionality.
- Every sigmoid transformer implements weighted loopy belief propagation on its implicit factor graph, verified formally.
- Transformers can implement exact belief propagation on any declared knowledge base, with provably correct probability estimates for non-circular dependencies.
- Uniqueness is proven: sigmoid transformers producing exact posteriors must have BP weights, with no alternative paths.
- The transformer layer's structure is delineated: attention as AND, FFN as OR, mirroring Pearl's gather/update algorithm.
- Experimental results corroborate the Bayesian network characterization, showing practical viability of loopy belief propagation.
- Verifiable inference requires a finite concept space; without grounding, correctness is undefined, making hallucination a structural consequence.
Preserving scientific integrity in academic publishing: Navigating artificial intelligence, journal policies, and the impact factor as a quality indicator - PubMed
a month ago
- Challenges to scientific integrity include AI integration, mega-journals, and impact factor manipulation.
- Threats are categorized into external pressures (AI misuse, metric-driven models) and internal systemic flaws ('publish or perish' culture, methodological fragility).
- Mega-journals improve accessibility and dissemination but may weaken peer review rigor due to high-volume publishing.
- Proposed solutions include AI transparency frameworks (e.g., CONSORT-AI) and redefining impact metrics to emphasize reproducibility and societal impact.
- Global cooperation (e.g., DORA, COPE) is essential to standardize ethics and address resource disparities.
- Encourages career opportunities based on publication quality over quantity to uphold scientific integrity.
INTELCAPE: A Deep Learning-Powered System for Automated, High-Accuracy Crohn's Disease Diagnosis via Capsule Endoscopy - PubMed
a month ago
- INTELCAPE is a deep learning-powered system for automated Crohn's disease (CD) diagnosis via capsule endoscopy (CE).
- The system integrates ResNet, Transformer, and EfficientNet architectures for hierarchical processing of CE videos.
- INTELCAPE achieved high accuracy in small-intestine segmentation (IoU scores up to 96.87%) and lesion detection (AUC up to 0.993).
- For CD diagnosis, INTELCAPE demonstrated robust generalizability with AUCs of 0.982 and 0.984, comparable to specialists but 10-fold faster.
- The system improved doctors' diagnostic accuracy from 76.7% to 94.8% and reduced interpretation time from 67.9 to 22.5 minutes.
- INTELCAPE is particularly beneficial for less-experienced clinicians, enhancing both accuracy and efficiency in CD diagnosis.
Identifying neuropathologic disease in primary progressive aphasia using narrative speech - PubMed
a month ago
- Artificial intelligence is applied to narrative speech to predict neuropathologic disease in primary progressive aphasia (PPA).
- Natural language processing extracts features from transcribed narratives of the Cinderella story.
- Machine learning models classify participants as normal controls (NC) or PPA with Alzheimer's disease (AD) or frontotemporal lobar degeneration (FTLD-4Rtau).
- Models successfully distinguish AD from FTLD-4Rtau and NC from PPA narratives.
- Feature permutation reveals different patterns of language domain and disease pathology contributions to classification.
- AI in naturalistic language tasks may aid in differential diagnosis and disease-specific interventions for PPA.
Artificial intelligence-based automated interpretation of images of electrocardiograms: development and multinational validation of ECG-GPT - PubMed
a month ago
- Development of ECG-GPT, an AI model for expert-level interpretation of ECG images.
- Validation across multinational datasets with 4.1 million ECGs, showing high diagnostic accuracy (AUROCs 0.80-0.96).
- ECG-GPT achieves a median pairwise similarity of 0.90 for diagnosis statements, outperforming baselines.
- Model performance consistent across diverse health settings, including US, Brazil, UK, and Germany.
- Potential to address ECG interpretation bottlenecks in low-resource settings.
- Authors disclose multiple patents, funding sources, and conflicts of interest related to AI and healthcare technologies.
Artificial Intelligence-Based Prediction of Invasion Depth in Colorectal Cancer via Endoscopic Imaging (With Video): A Narrative Review - PubMed
a month ago
- AI-based prediction of colorectal cancer (CRC) invasion depth via endoscopic imaging is crucial for treatment strategy.
- Recent advances include AI models using image-enhanced endoscopy, white-light imaging, and multi-modal data.
- High performance in diagnosing T1b CRC (AUC 0.851) was achieved, comparable to expert endoscopists.
- Challenges include limited sample sizes, dataset imbalance, and heterogeneity preventing pooled analysis.
- Large-scale, multicenter validation studies and open-access databases are essential for clinical implementation.
Reliable classification of polyps based on artificial intelligence: a development and validation study - PubMed
a month ago
- POLARIS is an AI-based tool developed to assist pathologists in classifying colorectal polyps, addressing the shortage of pathologists.
- The tool was trained using 15,079 whole-slide images from 2993 patients in the UK bowel cancer screening program (2014-2018).
- POLARIS categorizes polyps into five classes based on malignancy risk, regrouping them into two broader categories for clinical interpretation.
- External validation on 10,842 WSIs showed POLARIS correctly identified 98.94% of high-risk polyps (HGD and adenocarcinoma) and 83.04% of low-risk polyps (normal tissue and tubular adenomas with LGD).
- The model achieved a balanced accuracy of 86.65% in external validation and an AUROC of 0.9449 for distinguishing low-risk from high-risk polyps.
- Pathologists agreed with POLARIS over clinical diagnosis in 92.5% of reviewed cases where predictions differed.
- POLARIS has potential to improve diagnostic workflows by reducing the number of slides requiring pathologist review and highlighting high-risk regions.
- The development was funded by The Norwegian Cancer Society, with several authors having potential conflicts of interest related to commercialization.
Artificial intelligence in inflammatory bowel disease: bridging innovation, implementation and impact - PubMed
a month ago
- Artificial intelligence (AI) is transforming inflammatory bowel disease (IBD) management.
- Early AI applications focused on endoscopy, digital pathology, and imaging.
- Next-gen AI integrates multimodal data (endo-histo-omics) for deeper disease insights.
- AI applications include personalized treatment, remote monitoring, and decision support via large language models.
- Validated AI uses in IBD: diagnosis, monitoring, outcome prediction, and neoplasia surveillance.
- Key challenges: explainability, workflow integration, reimbursement, and environmental sustainability.
- Future of AI in IBD depends on ethical, collaborative, and transparent implementation.
AI boom risks widening wealth divide, says BlackRock's Larry Fink
a month ago
- AI boom risks widening inequality, benefiting only a few companies and investors.
- Larry Fink warns AI could exacerbate the wealth gap between rich and poor.
- Leading companies with resources to deploy AI at scale stand to gain disproportionately.
- Concerns of an AI investment bubble mirroring the dotcom crash conditions.
- Bank of England warns of potential 'sudden correction' in markets due to AI valuations.
- Fink suggests investing in stocks as an alternative to home ownership for wealth building.
- Rising housing costs and strict lending make home ownership less accessible and profitable.
- Encourages broader participation in capital markets to share in economic growth.
Generalist biological artificial intelligence in modeling the language of life - PubMed
a month ago
- Generalist biological artificial intelligence (GBAI) models the 'language of life' from DNA to cellular function.
- GBAI can interpret and generate DNA, RNA, proteins, and cellular systems.
- Potential applications include understanding disease pathways, advancing therapeutic design, and simulating biological activity.
- Challenges include data, biological complexity, scalability, and experimental validation.
- Synergies between language and structural AI, specialized models, and autonomous AI agents are key opportunities.
Oil at $150 will trigger global recession, says boss of financial BlackRock
a month ago
- Oil at $150 a barrel could trigger a global recession, warns BlackRock CEO Larry Fink.
- High oil prices due to Middle East tensions could lead to years of economic instability.
- Countries should diversify energy sources to ensure cheap energy and economic growth.
- Fink denies any similarities between current market conditions and the 2007-08 financial crisis.
- Investment in AI is not a bubble, but energy costs could hinder its expansion in the US and Europe.
- AI will create jobs in trades like plumbing and electrician work, shifting focus from traditional office jobs.
- There's a need to rebalance education and career paths to value technical training over university degrees.
Artificial Intelligence in hepatology: A position paper by the Italian Association for the Study of the Liver - PubMed
a month ago
- AI is transforming hepatology by improving diagnosis, prognosis, and treatment.
- The paper reviews AI applications in various liver diseases including MASLD/MASH, alcohol-related liver disease, and hepatocellular carcinoma.
- Key challenges include data quality, algorithmic fairness, explainability, and reproducibility.
- The AISF emphasizes the need for robust external validation and adherence to international reporting standards.
- A national roadmap is proposed for responsible AI adoption, focusing on ethics, transparency, and equity.
PNET-PRISM: a multicenter-validated radiomics nomogram for noninvasive grading of pancreatic neuroendocrine tumors - PubMed
a month ago
- PNET-PRISM is a radiomics nomogram for noninvasive grading of pancreatic neuroendocrine tumors (PNETs).
- The study included 407 patients across training, validation, and external test cohorts.
- PNET-PRISM integrates CT radiomics features from intratumoral, peritumoral, habitat, and deep learning domains.
- The model achieved high AUCs (0.92, 0.89, 0.87) in training, validation, and external test sets.
- PNET-PRISM showed 100% sensitivity in external validation and improved grading in cases where EUS-FNA failed.
- The M-DLR Score was associated with progression-free survival (HR = 2.050).
- The tool addresses limitations of EUS-FNA and aids in treatment selection when biopsy is inconclusive.
Artificial Intelligence in Cardiovascular Medicine: Focus on Hypertension - PubMed
a month ago
- Hypertension is a major global risk factor for cardiovascular disease, but blood pressure control rates are low despite available treatments.
- Current care models are limited due to episodic data, population-based approaches, and incomplete understanding of blood pressure regulation.
- AI enables integration of multimodal data and modeling of complex relationships, offering potential for predictive and personalized hypertension management.
- AI applications in hypertension include risk prediction, phenotyping, monitoring via wearables, clinical trial analysis, and precision medicine.
- Key requirements for effective AI include robust validation, interpretability, workflow integration, and equity considerations.
- Challenges include the limitations of new technologies like cuffless BP devices and ethical, regulatory, and implementation hurdles.

first prev13next

About|Login

#artificial intelligence

Gaming publisher's CEO used ChatGPT in failed bid to avoid paying $250M bonus

Beyond artificial intelligence psychosis: a functional typology of large language model-associated psychotic phenomena - PubMed

AI (2014)

Token-guided multimodal prognosis in hepatocellular carcinoma: a framework steered by tumour-stroma ratio - PubMed

Transformers Are Bayesian Networks

Preserving scientific integrity in academic publishing: Navigating artificial intelligence, journal policies, and the impact factor as a quality indicator - PubMed

INTELCAPE: A Deep Learning-Powered System for Automated, High-Accuracy Crohn's Disease Diagnosis via Capsule Endoscopy - PubMed

Identifying neuropathologic disease in primary progressive aphasia using narrative speech - PubMed

Artificial intelligence-based automated interpretation of images of electrocardiograms: development and multinational validation of ECG-GPT - PubMed

Artificial Intelligence-Based Prediction of Invasion Depth in Colorectal Cancer via Endoscopic Imaging (With Video): A Narrative Review - PubMed

Reliable classification of polyps based on artificial intelligence: a development and validation study - PubMed

Artificial intelligence in inflammatory bowel disease: bridging innovation, implementation and impact - PubMed

AI boom risks widening wealth divide, says BlackRock's Larry Fink

Generalist biological artificial intelligence in modeling the language of life - PubMed

Oil at $150 will trigger global recession, says boss of financial BlackRock

Artificial Intelligence in hepatology: A position paper by the Italian Association for the Study of the Liver - PubMed

PNET-PRISM: a multicenter-validated radiomics nomogram for noninvasive grading of pancreatic neuroendocrine tumors - PubMed

Artificial Intelligence in Cardiovascular Medicine: Focus on Hypertension - PubMed