Top 10 AI Highlights of 2023: A New Order of Large Models is Emerging
-
When future generations look back on history, they may refer to 2023 as the Year of AI. Large AI models have burst into our lives in a disruptive manner, advancing at an astonishing pace. From Q&A dialogues to programming assistance, from image analysis to autonomous creation, the capabilities demonstrated by AI have exceeded most people's expectations, leading many to exclaim: 'The future belongs to AI.'
With the release of ChatGPT, major global tech companies have accelerated the development of AI large models, bringing infinite possibilities to this emerging field. Next, we will review the top 10 AI large models of the past year in terms of innovation, application areas, and AI performance, allowing everyone to easily understand the forefront capabilities of AI large models today. (The ranking is in no particular order and follows alphabetical order.)
1. ChatGPT: The 'Pioneer' of AI Large Models
ChatGPT is hailed as the progenitor of AI large models. Developed by OpenAI and first released to the public in November 2022, as the first phenomenal AI model, ChatGPT has now evolved to version 4.0, offering processing capabilities far surpassing ChatGPT 3.5, along with features such as plugin loading and real-time internet connectivity.
Currently, people have designed thousands of different plugins for ChatGPT4.0, ranging from programming to drawing, from poetry writing to music composition. ChatGPT can be said to be proficient in everything. Of course, the ability to achieve these functions is mainly because ChatGPT itself is a multimodal AI model that can recognize and analyze various types of data.
ChatGPT is also currently the strongest AI model that most people can access.
2. Claude: OpenAI's 'Backyard'
Claude might be an unfamiliar name, but the team behind it is quite notable. Its founder, Dario Amodei, is the former Vice President of OpenAI, who led a group of employees to establish Anthropic. After a series of improvements and upgrades, Claude is now regarded as the AI model second only to ChatGPT, and it is particularly adept at processing long-form text.
Claude, after a recent upgrade, can now support up to 200k text input, far exceeding ChatGPT4.0 Turbo's 128K. However, compared to ChatGPT's active user ecosystem, Claude has a much smaller user base and application scope.
During the recent internal turmoil at OpenAI, Claude became a backup choice for many enterprises. Since most of Claude's core team members come from OpenAI, it can be considered a weaker version of ChatGPT to some extent. Although it is not as powerful as ChatGPT in overall performance, it has certain advantages in long-text reading and answering.
However, shortly after its release, Gemini became the subject of jokes when it claimed to be Baidu's Ernie Bot (in Chinese contexts) or OpenAI's ChatGPT (in English contexts) in its responses. Although Google later conducted emergency maintenance on Gemini to patch the loophole, it still triggered heated discussions among netizens.
Currently, Gemini is still in the testing phase. Whether Google's multimodal large model can surpass ChatGPT remains unknown. But at least from the user experience perspective, Gemini combines the advantages of many AI models under Google's umbrella, potentially bringing pleasant surprises in practical applications.
4. Huawei Cloud Pangu: The Leader in Full-Ecosystem AI
Huawei Cloud's Pangu AI model, though not widely known among the general public, has made the farthest and broadest strides in commercialization within the industry. The Pangu AI model includes a collection of multiple AI models such as NLP models, CV models, and scientific computing models, enabling its application across various industries including scientific research, education, transportation, and manufacturing.
To maximize the performance of large AI models, Huawei has integrated its industrial and product advantages to deploy the Pangu AI model in sectors such as mining, government, finance, and meteorology, with multiple applications already implemented. Additionally, the Pangu AI model is the first Chinese large-scale model with training parameters exceeding 100 billion, ranking among the top in overall capability and leading in industrial applications.
5.LLaMA2: The Popular 'Metaverse' Player
LLaMA2 is an AI model developed by the internet giant Meta. As an open-source AI model, LLaMA2 provides various AI models with parameter sizes ranging from 7B to 65B, serving users with different needs. Thanks to LLaMA2's outstanding efficiency, it still delivers impressive performance even with lower parameter counts, making it more advantageous for localized deployment on small and mobile devices.
Compared to LLaMA, LLaMA2 has enhanced multilingual processing capabilities while strengthening natural conversation skills and long-text comprehension. The generated content is more coherent, reducing the likelihood of errors. Combined with its open-source nature, this enables rapid adoption and widespread use.
In multiple tests, LLaMA2 has achieved top-tier results among open-source models, making it one of the most popular open-source AI models available. Although LLaMA2 still falls short of leading AI models like ChatGPT, it remains a solid choice for AI enthusiasts and small-to-medium enterprises.
6. Tongyi: User-Focused, Close to Reality
Tongyi is a series of Chinese AI models launched by Alibaba, and it is also one of the models familiar to Xiaolei. Currently, Tongyi includes multiple AI model portals such as Tongyi Qianwen, Tongyi Wanxiang, and Tongyi Lingma, which can help solve many problems. For example, Tongyi Qianwen is a natural conversation AI model that provides various text-based functionalities.
As Alibaba's Chinese AI model, Tongyi Qianwen has more advantages in querying business information and other data, with higher accuracy. Currently, Tongyi 2.0 has been upgraded to an AI model with hundreds of billions of parameters, demonstrating better performance in language understanding and problem-solving. Next, Alibaba plans to integrate it into apps like DingTalk to provide users with real-time online AI support, even enabling diverse functions such as AI-powered ordering.
Based on Alibaba's product ecosystem matrix, the Tongyi large model has been deployed in multiple apps and provides services to users. It can be said to be the domestic large model most closely connected with users. From DingTalk to Taobao, you can see the presence of the Tongyi large model. Interested friends might want to try Taobao's "Ask Taobao" feature, which offers many surprises.
7. Wenxin Yiyan: The Outstanding Chinese Large Model Even Google Needs to 'Learn' From
ERNIE Bot is a next-generation knowledge-enhanced large language model launched by Baidu. Benefiting from Baidu's long-term investment in AI technology and its leading position in the search engine market, ERNIE Bot is the fastest-growing AI large model in China. It delivers outstanding performance in various areas such as writing, creative copywriting, Q&A, consulting, and coding, while also providing content that better aligns with the Chinese context.
Currently, the ERNIE Bot web platform offers a rich variety of application scenarios and prompt templates in its "Treasure Box" feature, covering most of the daily needs of Chinese netizens. Additionally, it includes an extensive plugin marketplace with tools like PPT Assistant, Long Document Summary Q&A, Academic Search Assistant, and Mind Mapping, further expanding the capabilities of the large model.
It must be said that ERNIE Bot does an excellent job in understanding user needs, offering many interesting features. Currently, ERNIE Bot is available for free, based on the ERNIE 3.5 version. To experience the more powerful ERNIE 4.0 version, a membership is required. However, based on personal experience, the cost-performance ratio is quite high.
8. iFlytek Spark: The 'Translator' in the AI Field
iFlytek Spark, as the name suggests, is a product of iFlytek. As a leading artificial intelligence enterprise in China, iFlytek has significant advantages in speech synthesis and recognition, multilingual capabilities, and cognitive intelligence. After entering the field of AI models, it has fully leveraged its strengths, delivering excellent performance in voice conversations, translation, and copywriting. Based on the Spark large model, iFlytek is also integrating multiple hardware products, allowing users to carry and use the Spark model on the go.
Currently, the parameter scale of iFlytek Spark has exceeded 100 billion and outperforms ChatGPT in tests, though it still lags behind ChatGPT4.0. According to iFlytek's plan, the Spark AI large model is expected to reach the level of ChatGPT4.0 by the first half of 2024.
Additionally, iFlytek Spark has launched client applications such as Content Master to provide users with more direct and convenient solutions, aiding in better management of content publishing, creation, and review processes. With the help of AI, our work efficiency is being significantly enhanced.
9. Youdao Ziyue: The 'Purple Star' of Intelligent Education
Youdao Ziyue is a new-generation educational large model launched by NetEase. It is a vertical large model developed by Youdao specifically for educational scenarios and one of the first to receive national certification. NetEase Youdao has previously introduced six innovative applications based on this model, including large-model translation, essay guidance, grammar explanations, virtual oral coaches, and AIBox, which have been integrated into various software and hardware products such as Youdao Dictionary, Youdao Dictionary Pen, and Youdao Learning Machines.
Ziyue is a rare targeted vertical large model that performs exceptionally well in educational applications compared to general-purpose large models. This model incorporates advanced natural language processing technologies, making it particularly suitable for educational environments, where it can provide personalized learning experiences and enhance teaching effectiveness.
Through the integration of educational large models with software and hardware products, the application scenarios of Ziyue Large Model have been widely expanded, with its user experience surpassing general large models in many aspects. Compared to general large models, Ziyue can provide learners with personalized learning experiences in the education field, helping them achieve more efficient learning outcomes. With the increasing emphasis on online education, the development prospects of Ziyue Large Model are very promising.
10. Zhipu AI: The New Favorite in Mobile and Automotive Fields
Although Zhipu AI is not as well-known as models like Tongyi, Wenxin, and Spark, its team was one of the earliest forces to enter the AI model field. Zhipu AI released its self-developed GLM model with billions of parameters as early as 2021 and launched the third-generation base model, ChatGLM3, in October this year, providing the industry with 1.5B and 3B small AI models that can be quickly deployed on mobile and even automotive platforms.
As one of the earliest open-source AI models in China, Zhipu AI has contributed significantly to the development of domestic AI models and remains a leader in the field. Compared to other AI models, Zhipu AI excels in efficiency and other aspects, and its open-source nature has made it a preferred partner for many small and medium-sized enterprises in China.
Over the past year, the development and application of AI large models have reached new heights. They have not only made significant technological progress but also demonstrated broad potential across multiple fields.
From the versatility of ChatGPT to Claude's long-text processing, Falcon's open-source features, and Huawei's Pangu's commercial applications, each model has contributed to the advancement and popularization of AI in its unique way. Whether it's Google's Gemini, Alibaba's Tongyi, Baidu's Wenxin Yiyan, iFlytek's Spark, or Zhipu AI, these AI large models are continuously pushing the boundaries of technology, enhancing the efficiency and accessibility of artificial intelligence.
The development of AI in 2023 marks the beginning of a new era. These large AI models not only represent the cutting edge of technology but also indicate the direction of future development. As these models continue to evolve and expand their applications, we can expect AI to play a key role in more fields, bringing us more innovation and convenience.