Foundational models such as ChatGPT critically depend on vast data scales the internet uniquely enables. This implies exposure to material varying widely in logical sense, factual fidelity, moral value, and even legal status. Whereas data scaling is a technical challenge, soluble with greater computational resource, complex semantic filtering cannot be performed reliably without human intervention: the self-supervision that makes foundational models possible at least in part presupposes the abilities they seek to acquire. This unavoidably introduces the need for large-scale human supervision-not just of training input but also model output-and imbues any model with subjectivity reflecting the beliefs of its creator. The pressure to minimize the cost of the former is in direct conflict with the pressure to maximise the quality of the latter. Moreover, it is unclear how complex semantics, especially in the realm of the moral, could ever be reduced to an objective function any machine could plausibly maximise. We suggest the development of foundational models necessitates urgent innovation in quantitative ethics and outline possible avenues for its realisation.
Download full-text PDF |
Source |
|---|---|
| http://dx.doi.org/10.1007/s00429-023-02662-7 | DOI Listing |
Publication Analysis
Top Keywords
Similar Publications
Elite Alleles of EPE1 Identified via Genome-wide Association Studies Increase Panicle Elongation Length in Rice.
Rice (N Y)
January 2025
College of Agronomy, Anhui Agricultural University, Hefei, 230000, China.
Panicle elongation length (PEL), which determines panicle exsertion, is an important outcrossing-related trait. Mining genes controlling PEL in rice (Oryza sativa L.) has great practical significance in breeding cytoplasmic male sterility (CMS) lines with increased PEL and simplified, high-efficiency seed production.
View Article and Find Full Text PDFPreclinical evaluation of zirconium-89 labeled anti-Trop2 antibody-drug conjugate (Trodelvy) for imaging in gastric cancer and triple-negative breast cancer.
Eur J Nucl Med Mol Imaging
January 2025
Departments of Radiology and Medical Physics, University of Wisconsin - Madison, Madison, WI, 53705, USA.
Purpose: Trophoblast cell-surface antigen 2 (Trop2) is overexpressed in various solid tumors and contributes to tumor progression, while its expression remains low in normal tissues. Trop2-targeting antibody-drug conjugate (ADC), sacituzumab govitecan-hziy (Trodelvy), has shown efficacy in targeting this antigen. Leveraging the enhanced specificity of ADCs, we conducted the first immunoPET imaging study of Trop2 expression in gastric cancer (GC) and triple-negative breast cancer (TNBC) models using Zr-labeled Trodelvy ([Zr]Zr-DFO-Trodelvy).
View Article and Find Full Text PDFGenerative artificial intelligence enables the generation of bone scintigraphy images and improves generalization of deep learning models in data-constrained environments.
Eur J Nucl Med Mol Imaging
January 2025
Department of Biomedical Imaging and Image-guided Therapy, Division of Nuclear Medicine, Medical University of Vienna, Spitalgasse 23, Vienna, 1090, Austria.
Purpose: Advancements of deep learning in medical imaging are often constrained by the limited availability of large, annotated datasets, resulting in underperforming models when deployed under real-world conditions. This study investigated a generative artificial intelligence (AI) approach to create synthetic medical images taking the example of bone scintigraphy scans, to increase the data diversity of small-scale datasets for more effective model training and improved generalization.
Methods: We trained a generative model on Tc-bone scintigraphy scans from 9,170 patients in one center to generate high-quality and fully anonymized annotated scans of patients representing two distinct disease patterns: abnormal uptake indicative of (i) bone metastases and (ii) cardiac uptake indicative of cardiac amyloidosis.
2-dodecyl-6-methoxycyclohexa-2,5-diene-1,4-dione protects against MPP-induced neurotoxicity by ameliorating oxidative stress, apoptosis and autophagy in SH-SY5Y cells.
Metab Brain Dis
January 2025
Key Laboratory of Longevity and Aging-Related Disease of Chinese Ministry of Education, Center for Translational Medicine, School of Basic Medical Sciences, Guangxi Medical University, Nanning, Guangxi, China.
2-dodecyl-6-methoxycyclohexa-2,5-diene-1,4-dione (DMDD) is a cyclohexanedione compound extracted from the roots of Averrhoa carambola L. Several studies have documented its beneficial effects on diabetes, Alzheimer's disease, and cancer. However, its potential neuroprotective effects on Parkinson's disease (PD) have not yet been explored.
View Article and Find Full Text PDFBright and Specific Targeting of Metastatic Lymph Nodes in Orthotopic Mouse Models of Gastric Cancer with a Fluorescent Anti-CEA Antibody.
Ann Surg Oncol
January 2025
Department of Surgery, University of California San Diego, La Jolla, CA, USA.
Background: Gastric cancer poses a major diagnostic and therapeutic challenge. Improved visualization of tumor margins and lymph node metastases with tumor-specific fluorescent markers could improve outcomes.
Methods: To establish orthotopic models of gastric cancer, one million cells of the human gastric cancer cell line, MKN45, were suspended in 50 μl of equal parts PBS and Matrigel and injected into the nude mouse stomach with a 29-gauge needle.
Want AI Summaries of new PubMed Abstracts delivered to your In-box?
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!