A new wave of AI language models is emerging and competing with OpenAI’s ChatGPT.
As the field of artificial intelligence (AI) and natural language processing (NLP) continues to advance, a new generation of AI language models is emerging to challenge the dominance of popular models like OpenAI’s Chat Generative Pre-trained Transformer, or ChatGPT. To get a piece of the market, various companies, including large tech companies like Google and Baidu, along with smaller startups and research organizations, are joining the frenzy and developing new language models. Some are even designed for specific tasks or industries, pushing the boundaries of NLP.
As these models become increasingly advanced, they’re capable of tackling challenges that were once thought impossible. Let’s take a look at the capabilities of OpenAI’s ChatGPT and the emerging challengers that are on the path to compete with it.
A brief introduction to ChatGPT
ChatGPT is a conversational AI chatbot created by OpenAI, an AI research company based in San Francisco co-founded by Sam Altman and Elon Musk. Launched in November 2022, the chatbot is built on the foundation of GPT-3.5, a deep learning-based language model that creates text that resembles human writing. In just five days after its release, ChatGPT has reached more than one million users. Having already acquired 100 million monthly users, and it’s growing faster than TikTok.
Unlike its predecessor, GPT-3, ChatGPT takes the concept of language modeling to the next level with its ability to generate more elaborate text and even compose poems. It also boasts a remarkable feature—the capacity for recall. The chatbot can retain previous interactions and reference them during a conversation, making it a more interactive and engaging experience. A chatbot that uses OpenAI’s GPT-3 language model is Replika. Many are using it as an AI friend or romantic partner for how it can generate output that makes users feel like they’re talking to a real person.
Currently, ChatGPT is freely available to the public as a research preview. Users can experience the chatbot’s capabilities by accessing the dedicated website. The company has recently launched a paid version of ChatGPT called ChatGPT Plus for a monthly subscription fee of US$20. This new version offers several benefits for its subscribers, including general access to the chatbot at all times, even during peak usage hours, faster response times and priority access to new features and improvements.
Meet the new contenders
In January, Microsoft, which has a strategic partnership with OpenAI, announced a US$1 billion investment in the company, followed by revealing the integration of OpenAI’s GPT-4 model with its search engine Bing on February 7. To remain competitive, many big tech companies also revealed their plans to roll out their AI soon. Let’s explore the new generation of AI language models and see how they stack up against the reigning champion, ChatGPT.
On February 6, Google announced that its AI chatbot, Bard, is set to launch in the coming weeks—likely a response to the perceived threat that ChatGPT poses to Google’s search business. With the emergence of ChatGPT, the tech giant has declared a “code red”, indicating that Google is taking the competition seriously.
As described in a recent blog post by Google CEO Sundar Pichai, Google Bard is a language model that competes with ChatGPT and utilizes Google’s LaMDA (Language Model for Dialogue Applications) to generate responses, harnessing the vast amount of fresh information available on the internet. This sets it apart from ChatGPT, which is limited to information up until 2021 and lacks access to online resources. The conversational chatbot uses information from the internet to generate new and accurate responses.
Like ChatGPT, Bard is built on the Transformer architecture, which was developed by Google and made available to the public as an open-source project in 2017. The Transformer technology can make predictions based on input and is widely utilized in NLP and computer vision.
Despite Google’s plans to release Bard to a wider audience in the near future, the tech giant has experienced a few setbacks in the process. For instance, during a recent press conference held in Paris, a presenter for Google was unable to proceed with a demo due to a misplaced phone. Additionally, the chatbot made a factual error during its initial demonstration, causing experts to question its reliability. These incidents resulted in a loss of US$100 million in market value for Google’s parent company, Alphabet.
Despite these challenges, Google and Alphabet remain committed to improving the technology and bringing Bard to a wider audience. Additionally, the company intends to integrate the chatbot into its search engine, so users can interact with Bard directly when searching on Google.
Baidu’s Ernie Bot
Ernie is an AI chatbot project developed by Chinese tech giant Baidu. According to a spokesperson who spoke to CNN, the chatbot, known as “Wenxin Yiyan” in Chinese or “Ernie Bot” in English, is currently undergoing internal testing and will be launched in March 2023. Ernie, an acronym for “Enhanced Representation through Knowledge Integration”, is built on a language model Baidu developed in 2019.
As per Reuters, Ernie has been evolving since 2019 and now can undertake various tasks, such as language understanding, language generation and text-to-image generation. A Baidu spokesperson said to The Register, “What sets Ernie apart from other language models is its ability to integrate extensive knowledge with massive data, resulting in exceptional understanding and generation capabilities.”
Due to internet censorship policies in China, ChatGPT and essential Google services are not easily accessible to Chinese users. Hence, Baidu’s Ernie bot might be able to capture the Chinese market, given that it is specifically designed to work within China’s internet environment. Additionally, as Baidu is a dominant player in the Chinese technology sector, it has access to a wealth of user data, which Ernie can leverage to continually improve its performance.
Other players in the race
Chinese e-commerce giant Alibaba is the latest entrant in the race to develop a ChatGPT-style AI chatbot. A spokesperson from the company stated that Alibaba is currently conducting internal tests on their AI technology. Although the company has been working on generative AI since 2017, there is no set timeline for the release of their ChatGPT competitor to the market.
Amazon is another major player that is reportedly working on a ChatGPT-like technology. During a recent announcement, CEO Andy Jassy confirmed that the company has been developing generative AI technology that shares similarities with the ChatGPT model. While details regarding the technology’s scope and release timeline were not disclosed, the announcement underscores Amazon’s ongoing efforts to enhance its capabilities in the AI space.
The way forward
The chatbot mega-race is a highly competitive space, with several companies investing significant resources into developing a more sophisticated chatbot. While each company is pursuing a different strategy to achieve its goals, the ultimate aim is to create chatbots that can understand and respond to natural language queries more accurately and efficiently. It will be interesting to see how this space evolves and which company will emerge as the leader in the coming years.
- 5 Essential Reasons Chatbots Fail—and Will ChatGPT, Too?
- Can AI Help You Flirt Better?
- Anima vs. Replika: Reviewing Top AI Boyfriends on Google Play Store
- Is Using AI for Academic Writing Cheating?
Header image courtesy of Pexels