Google AI’s MDA versus OpenAI’s ChatGPT

[ad_1]

The latest innovation in conversational technology, ChatGPT, has won all the awards. Since ChatGPT, an AI-powered text-based chatbot was made available by OpenAI in November, it has become quite popular. However, a competitor that entered the market before OpenAI’s run with an AI chatbot and which many still feel is the superior of the two deserves far more attention. Some Google employees are reportedly wondering if they missed an opportunity given the company’s long history of bragging about its AI proficiency. Analysts are also speculating that AI chatbots could one day challenge Google’s hegemony.

LaMDA vs. Chat GPT

The Language Model for Dialogue Applications (LaMDA) is a Transformer-based neural language model that is pre-trained using 1.56T words of conversation data and freely accessible online pages. By gathering feedback from the pre-trained model, the refined model, and human evaluators, the progress of LaMDA is measured. It has up to 137B parameters. The model is also adapted according to the criteria of Quality, Safety and Fundamentality.

With 175B parameters, ChatGPT is based on the GPT-3.5 architecture. The GPT-3.5 series includes three models: text-davinci-002, trained by supervised tuning on human written demonstration, and samples rated 7/7 by human labelers on overall quality scores and the newer text-davinci- 003. This updated version includes Human Feedback Reinforcement Learning (RLHF), a reward-based model trained on comparisons by humans. Code-davinci-002 is the base template for code completion tasks. LaMDA-like supervised knowledge is used by ChatGPT, where human AI trainers can use model recommendations to create responses and train the model by simulating both the user and the AI ​​assistant.

Market rumors state that LaMDA was never released because the conversational AI chatbot platform didn’t work with Google’s current advertising strategy, which generated 81% of its revenue in fiscal 2021. As it is known that the AI ​​chatbot imitates human speech from the Internet and spreads some types of hate speech, fake news, racist and sexist statements, and a high degree of informational inaccuracy, there may also be reputational risks. The LaMDA AI Platform is not currently accessible to the general public and is only accessible to a select number of AI developers through the AI ​​Test Kitchen.

Undoubtedly, Google needs to revamp its Google search engine with AI technology to stay profitable from advertising while keeping pace with fierce competition and changing customer needs. It is also evident that human oversight is needed to ensure that the data offered on the platform is true, reliable and safe for the general public due to the regulatory struggles in different nations over the years. Depending on how events play out in the coming months, the company’s defensive actions against ChatGPT could also result in increased R&D spending.

Google’s search engine comes free, with the platform receiving money from advertisements. The Google Search & Others division generated $119.84 billion in revenue for the company in the first nine months of 2022, an increase of 13.43% year over year.

On the other hand, LaMDA has an edge in this situation due to the different metrics it generates in its responses. For example, the rooting metric validates solutions based on trusted external sources. Similarly, the Quality Gauge evaluates responses based on sensitivity, specificity, and interest (SSI) criteria. In other words, it ensures that the answers are not generic, make sense in the context of the question, and are also wise, surprising, or funny.

Summing up –

Meta has increasingly kept information about its chatbots under wraps. With the introduction of ChatGPT, anticipation that Google will soon incorporate the newest iteration of the model into its search engine is growing. Google also controls the search market. A few months ago the third version of Blenderbot was published, the Meta chatbot that competes with Google and OpenAI. The conversational AI prototype has long-term memory and is based on 175B metrics. The model generates output using the internet, memory and previous dialogue. An announcement can be anticipated by Meta and Google when their chatbots are fully prepared, especially in light of Google’s struggles the last time they made a premature release.

Due to the hype around ChatGPT, Google is under pressure to develop and adapt quickly or risk falling behind competitors like TikTok and Meta. Since ChatGPT uses the conversational AI platform created by Google engineers in 2017, it’s time for the company to launch LaMDA. The winner is still up for grabs, but one thing is for sure: Google is more than just a chatbot demo.

Don’t forget to join our Reddit page and Discord channel, where we share the latest AI research news, cool AI projects, and more.

Dhanshree Shenwai is a consulting content writer at MarktechPost. He is a software engineer and works as a Delivery Manager in a major global bank. He has good experience in FinTech companies covering finance, cards and payments and banking with a keen interest in AI applications. He is enthusiastic about exploring new technologies and advancements in today’s changing world.

Meet Hailo-8: An AI Processor Using Computer Vision for Multi-Person Multi-Camera Re-Identification (Sponsored)

[ad_2]