关于人工智能大模型发展的思考与展望

Thoughts on and Prospects of the Development of I arge Model of Artificial Intelligence

当前,大模型的发展速度日益加快,且其发展的道路和方向已成为人工智能发展的重要命题。如何让机器具有智力,成为人类自身发展的好伙伴、好帮手,是大模型发展面临的新挑战。我们认为,未来大模型将从表述内容的“搜索范式”进化到阐述论证观点的“价值范式”,建立自己的价值观。在这样的价值范式下,机器的人性化不断增强,创造力得到进一步发展,从而使幻思的能力成为合乎世界模型的有意义的创造,最终推动大模型步入“创造范式”,使人与机器的共生、共存、共创成为可能,形成崭新的社会形态。为此,我们需要建构面向未来的技术理想,对人工智能的发展及其推动人类发展进步的前景充满信心,不畏试错、大胆创新,走出一条我们自己的大模型道路。

At present,the development of large models is accelerating,and the paths and directions of their development have become significant topics in the advancement of artificial intelligence.How to make a machine to become intelligent and be a good partner and helper of human development is a new challenge facing the development of large model.We propose that future large model will evolve from a"search paradigm"that expresses content to a"value paradigm"that articulates arguments,establishing its own values.Under such a value paradigm,humanization of machine is continuously enhanced,and creativity is further developed,so that the ability of illusion becomes a meaningful creation in line with the world model,and finally promotes large model to enter the"creation paradigm".Furthermore,symbiosis,coexistence and co-creation of people and machines will become possible,forming a new social form.To this end,we need to construct forward-looking technical ideals,have full confidence in the development of artificial intlligence and its prospects for promoting human development and progress,be fearless of trial and error and be bold in innovation,and walk out of a large model development path of our own.