How predictable is human life? Recent study Results of a study conducted by researchers at the Technical University of Denmark and Northeastern University suggest that artificial intelligence (AI) Transformer models can predict major human life events, including death.
“Our framework allows researchers to identify new potential mechanisms that influence life outcomes and the potential for associated personalized interventions,” said lead author Denmark University of Technology professor Sune Lehmann and co-authors Germans Savusissens, Tina Eliassi-Rudd and Lars wrote: Kai Hansen, Laust Huvas Mortensen, Rau Lilleholt, Anna Rogers, Ingo Zettler.
Researchers report that their proof-of-concept AI model shows a high degree of predictive accuracy. In the development of artificial intelligence machine learning, two important components that influence model accuracy are the algorithms used and the datasets.
In machine learning, the quality of an AI algorithm depends on the depth and breadth of the training data. To put this in context, ChatGPT (Chat Generative Pretrained Transformer), a large-scale language model (LLM) developed by OpenAI, is a powerful AI chatbot trained on large amounts of data. GPT-3 has 175 billion parameters and uses a large amount of internet data, including 570 gigabytes from 2016 to 2019 from Common Crawl, an open repository of web crawl data, the WebText2 dataset, and data from English Wikipedia. was trained. According to the 2020 OpenAI preprint, two internet-based book datasets called Books1 and Books2.
To train the AI Transformer model, the researchers used a large dataset containing individual-level details from a work and health database of 6 million Danish residents spanning several decades. The data included not only detailed information about life events, but also daily solutions with data on education, work, working hours, income, and health.
“We were able to observe how an individual’s life evolves in the face of different types of events (information about heart attacks mixed with information about salary increases and migration from urban to rural areas). ,” the researchers wrote.
The AI deep learning model used is called ‘life2vec’ and is based on the Transformer architecture. transformer The model was published in the paper “Attending Is All You Need” by Google researchers Ashish Vaswani, Illia Polosukhin, Jakob Uszhoreit, Noam Shazeer, Niki Parmar, Llion Jones, Lukasz Kaiser and colleagues at the 31st Conference on Neural Information Processing Systems in 2017. It was introduced in Aidan Gomez of the University of Toronto. Transformer models are widely used for purposes such as natural language processing (NLP), computer vision, and speech recognition.
In this current study, the researchers created life2vec using a design based on the BERT model. The BERT model stands for Bidirectional Encoder Representations from Transformers. BERT is an open source AI transformer released by Google in his 2018 for natural language processing.
“Our model allows us to predict a variety of outcomes, from early mortality to personality nuances, significantly outperforming state-of-the-art models,” the scientists report .
Through validation of their research prototype, it is clear that AI has the ability to predict human life events using powerful transducers and vast amounts of training data, and in the future will be used in health research and social sciences. can provide valuable insights.
Copyright © 2024 Kami Rosso All rights reserved.