By Dhanshree Shripad Shenwai
The most current innovation in conversational technology, ChatGPT, has won all the awards. Since ChatGPT, a text-based chatbot with artificial intelligence was made available by OpenAI in November, it has been quite popular. However, a competitor who entered the market before OpenAI’s run with an AI chatbot—and who many still feel to be the superior of the two—deserves a lot more attention. Some Google employees are reportedly questioning if they lost a chance given the company’s long history of boasting about its AI competence. Analysts are also speculating as to whether AI chatbots may one day challenge Google’s hegemony.
LaMDA Vs. ChatGPT
The Language Model for Dialog Applications (LaMDA) is a Transformer-based neural language model pre-trained using 1.56T words of freely accessible conversation data and online pages. By gathering feedback from the pre-trained model, fine-tuned model, and human raters, LaMDA’s progress is measured. It has up to 137B parameters. The model is also adjusted based on the criteria of Quality, Safety, and Groundedness.
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 fine-tuning on human-written demonstration, and samples rated 7/7 by human labelers on overall quality scores and the most recent text-davinci-003. This updated version includes reinforcement learning with human feedback (RLHF), a reward-based model trained on comparisons by humans. Code-davinci-002 is the base model for code completion tasks. LaMDA-like supervised knowledge is used by ChatGPT, where human AI trainers may utilize model recommendations to create answers and train the model while simulating both the user and the AI assistant.
Market rumors claim that LaMDA was never released because the conversational platform of the AI chatbot did not work with Google’s current advertising strategy, which generated 81% of its income in FY2021. Since the AI chatbot has been known to mimic human speech from the internet and spread some types of hate speech, fake news, racist and sexist statements, and a high degree of informative inaccuracy, there may also be reputational hazards. The LaMDA AI platform is presently not accessible to the general public and is only accessible to a select number of AI developers via the AI Test Kitchen.
Undoubtedly, Google has to revamp its Google Search Engine with AI technology to remain profitable from advertising while keeping up with the fierce competition and changing customer requirements. It is also evident that human control is required to guarantee that the data offered on the platform is true, trustworthy, and secure for the general public due to its regulatory struggles in several nations over the years. Depending on how events play out over the next months, the company’s defensive actions against ChatGPT may also result in rising R&D spending.
The Google Search Engine is provided without charge, with the platform receiving money from advertisements. The Google Search & Others division generated $119.84B in revenue for the firm in the first nine months of 2022, up 13.43% YoY.
On the other hand, LaMDA has an advantage in this situation due to the different metrics it generates in its answers. For instance, the groundedness metric validates the solutions based on reliable outside sources. Similarly, the quality meter evaluates replies according to sensibleness, specificity, and interest (SSI) criteria. In other words, it ensures the answers are non-generic, make sense in the context of the question, and are also wise, surprising, or humorous.
To sum it up –
Meta has been keeping information about its chatbots increasingly under wraps. With the introduction of ChatGPT, anticipation that Google would soon incorporate the most recent iteration of the model into its search engine is rising. Google also controls the search market. A few months ago, the third version of Blenderbot, Meta’s chatbot that competes with Google and OpenAI, was published. The conversational AI prototype has long-term memory and is based on 175B parameters. The model generates output using the internet, memory, and previous dialogue. An announcement can be anticipated from Meta and Google when their chatbots are fully prepared—especially in light of Google’s difficulties the last time they made a premature release.
Because of the buzz around ChatGPT, Google is under pressure to develop and adapt quickly or risk falling behind competitors like TikTok and Meta. Given that ChatGPT uses the conversational AI platform created by Google’s engineers in 2017, it is time for the company to launch LaMDA. The winner is still up in the air, but one thing is sure: Google is much more than simply a chatbot demonstration.
Dhanshree Shenwai is a Consulting Content Writer at MarktechPost. She is a Computer Science Engineer and working as a Delivery Manager in leading global bank. She has a good experience in FinTech companies covering Financial, Cards & Payments and Banking domain with keen interest in applications of AI. She is enthusiastic about exploring new technologies and advancements in today’s evolving world. (Source)