近日,艾伦人工智能研究院创始CEO、现董事会成员兼顾问奥伦·伊奇奥尼博士在接受每经网的专访时,对未来的人工智能发展趋势和成本变化进行了深入分析。

据伊奇奥尼博士表示,2024年将是多模态大模型的研发关键时期。传统的AI主要依靠单一模态的数据进行学习,而多模态大模型则可以同时处理多种类型的数据,如文本、图像和语音等。这种新型的AI不仅可以更好地理解和处理复杂的信息,还能实现更广泛的应用场景,例如自动驾驶、医疗诊断等。

除了技术发展外,伊奇奥尼博士还强调了AI成本的变化趋势。他引用摩尔定律指出,在AI时代,芯片的发展速度并不会因为技术难度的增加而放缓,反而会更加迅速。这意味着AI的训练和推理成本可能会以每18个月减半的速度持续下降,为AI技术的大规模应用提供更强的支持。

此次访谈为我们揭示了未来AI发展的新方向和潜力,也让我们看到了AI成本下降带来的巨大机遇。相信在不远的将来,AI将在更多的领域展现出其强大的力量。

英语如下:

News Title: “AI Giant Predicts: Multimodal Era is Coming, Costs Halved Every 18 Months!”

Keywords: AI development, multimodal, cost reduction

News Content:

Recently, Dr. Oren Etzioni, the founder and current board member and advisor of the Allen Institute for Artificial Intelligence, gave an in-depth analysis of the future trends of artificial intelligence development and cost changes during an exclusive interview with Netease Finance.

According to Dr. Etzioni, 2024 will be a key period for the development of large-scale multimodal models. Traditional AI mainly relies on single-modal data for learning, while multimodal large models can handle multiple types of data at the same time, such as text, images, and speech. This new type of AI can not only better understand and process complex information but also achieve broader application scenarios, such as autonomous driving and medical diagnosis.

In addition to technological development, Dr. Etzioni also emphasized the trend of change in AI costs. He cited Moore’s Law to point out that in the age of AI, the development speed of chips will not slow down due to the increase in technical difficulty, but will instead accelerate even more rapidly. This means that the training and inference costs of AI may decrease by half every 18 months, providing stronger support for the大规模应用 of AI technology.

This interview has revealed new directions and potential for the future development of AI, and also shown us the huge opportunities brought about by the decline in AI costs. We believe that in the near future, AI will demonstrate its powerful strength in more fields.

【来源】https://www.nbd.com.cn/articles/2023-12-26/3181902.html

Views: 2

发表回复

您的电子邮箱地址不会被公开。 必填项已用 * 标注