In today's deeply technologized world, search engines, such as Google, Yahoo, Bing, etc., have become essential to our daily lives. According to Statista, Google remains the leading search engine globally, with a desktop market share of around 84.69%. 

However, as search continues to evolve, it becomes increasingly clear that the current model has limitations. Users often struggle with biased or inaccurate search results. For instance, the search engine algorithm may unwittingly promote untested/unproven treatments or medical advice. More importantly, with the introduction of AI-powered chatbots, such as ChatGPT, there is a growing demand for more personalized and intuitive search experiences.

To meet this demand, Google has also released its own generative AI chatbot - Bard. Moreover, several other digital giants, such as Naver and Yandex, have announced their development of AI-powered conversational search products, SearchGPT and YaLM 2.0, all set to launch in 2023.

As the battle for chatbot supremacy heats up, we wonder what the future of search will look like. Will chatbots become the go-to interface for accessing information online? And how will generative AI continue to shape the search landscape?

The article explores the working of Bard, its limitations, and the future of chat-based search.

Conversational Search

Before we discuss Bard and its working, let’s explore the concept of conversational search. 

Conversational search is becoming wildly popular due to the ease voice assistants and chatbots offer, which use natural language processing to provide more human-like user experiences.

“The estimated size of the Global Conversational AI Market is expected to increase from $6.8 billion in 2021 to $18.4 billion by 2026, with a projected compound annual growth rate (CAGR) of 21.8% during the forecast period.” ~ MarketsandMarkets

‍

So what is conversational search?

At its core, the concept of 'conversational search' is getting hold of information in a way that mimics a natural conversation. This involves using natural language queries and receiving conversational responses rather than a list of links or a static webpage. 

Advanced techniques like natural language processing (NLP), text analysis, automatic speech recognition (ASR), and computer vision are employed in conversational search. They help machines make sense of user queries, understand intent, and provide the most relevant and accurate results.  

‍

What is Bard?

Google has developed a new chatbot called Bard, designed to respond to user queries and requests in a conversational manner. Like OpenAI's ChatGPT, Bard uses AI technology to generate high-quality responses based on online information. 

While Bard is not widely available for use yet, it is anticipated that Google Search will eventually integrate it so that users can access it through the search bar. Bard's capacity to mimic human dialogue will give users a more tailored and interesting search experience.

How Does It Work?

Google Bard is the latest chatbot that uses an experimental technology called LaMDA (Language Model for Dialogue Applications). LaMDA is a language model built on Transformer, the neural network architecture behind several generative AI use cases, such as OpenAI's ChatGPT's GPT-3 language model.

Google Bard is a lightweight version of LaMDA, designed to scale up for a broader audience and accumulate more feedback. It consumes significantly less computational power while supporting Google's internal testing activities.

LaMDA uses a statistical approach that employs the past words in a sequence to forecast the next ones. Unlike traditional task-based responses, LaMDA's advanced methodology permits a more natural dialogue flow, enabling users to switch between topics effortlessly. This technique is supported by concepts such as multimodal user intent, reinforcement learning, and suggestions to facilitate a more fluid conversation.

Bard’s Value Proposition

With Google's Bard, you can ask a highly contextualized question and receive a personalized response from its vast knowledge base. Unlike traditional search engines that provide straightforward answers, Bard can answer more abstract and open-ended queries such as "Which is easier to learn, the piano or guitar?" with detailed and nuanced responses. 

How does Bard do this? It draws data from its language model, which functions like a human brain, and from the web. In addition to collecting factual information, Bard is capable of analyzing and interpreting opinions expressed in various forms of text-based content, such as blogs and articles. As a result, it can provide more detailed and informative responses to a wide range of queries.

How Chatbots Improve Search?

While there are several limitations in current search, chatbots like ChatGPT and Google Bard can improve search in several ways: 

‍

1. Conversational search: Chatbots use NLP and machine learning processes to interact with users naturally. This helps make the search experience more conversational. 
2. Personalization: Since chatbots can learn from previous user interactions, they can personalize the search results based on user preferences and behavior.
3. Speed & efficiency: Unlike traditional search, chatbots can provide relevant answers quickly to users, reducing the time and effort required to find the desired information. 
4. Natural language understanding: Since chatbots like ChatGPT and Bard can understand natural language, they can help users find relevant information without the need for specific keywords or phrases. 

‍

Current Limitations of Search

While search engines have come a long way in terms of capabilities, they have some limitations that increase the significance of conversational search and AI chatbots. 

1. Website rankings in search engines are sometimes based on concepts such as intrinsic authority, which is often flawed and can be manipulated to influence rankings.
2. Search engines are typically biased towards established websites or paid content, with little attention to new, up-and-coming websites.
3. Even with the topic layer, search engines have limited capacity to understand the context and nuances of user queries, often providing results that are not relevant or accurate.
4. Search engines, such as Google, sometimes may have no choice but to display sites with inaccurate information if no alternative search results are available. 
5. Search engine rankings may be manipulated by commercial interests, including reliance on paid advertisements and the promotion of products.
6. Search engines leverage users’ lack of privacy measures to create profiles based on their interests, affecting the search results they see.

The Limitations and Challenges of AI Chat-Based Search

While AI chatbots and conversational search tools like Bard are changing how we search for information, they are not without limitations. For instance, Google Bard shared inaccurate information in its promotional video, which cost the company $100bn in share price value.

Some limitations and challenges of AI chat-based search are:

1. Lack of human context: AI chat-based search has limitations when it comes to responding to human queries, as they can only provide answers based on their training data. If a customer poses a question that exceeds the machine's capabilities, it can lead to a lack of comprehension and an inability to respond.
2. Training data dependent: Since chatbots depend on their training data, they can show inaccurate results if the data is biased or unreliable. However, chatbots like ChatGPT are developed to learn from past interactions and improve in the future.
3. Repetition in rule-based chatbots: Rule-based chatbots are trained to provide standard responses to user queries. Suppose a user asks a question that the chatbot hasn't been specifically trained to answer or rephrases the question. In that case, a chatbot may still provide the same answer, lacking the flexibility and adaptability of cognitive technology.
4. Privacy & security issues with AI chatbots: Chatbots used by businesses are often vulnerable to various security and privacy risks, including data protection issues where they collect and retain personal data without clear understanding. Furthermore, unencrypted conversations can expose users to third-party access and data breaches.
5. Language barriers: Language barriers can arise due to differences in dialect, regional variations, or colloquial expressions used by users. Most AI chatbots are trained in standard English or other commonly spoken languages, which may not account for the nuances of specific dialects or colloquial expressions used by different populations.
6. Fake news generation: The ability of chatbots to create and disseminate content poses a significant challenge in preventing the spread of fake news and misinformation, which can have serious consequences if it reaches a large audience. As chatbots are designed to mimic human conversation, they can be used to create false narratives or propagate misinformation.

Are AI Chatbots Humanity’s New Co-pilots?

Conversational AI has emerged as a valuable tool across various industries, serving as a copilot to support and enhance human capabilities. For example, 

1. In retail, chatbots can be available 24/7, handling customer inquiries and meeting their demands even during holidays. This expands the scope of customer support beyond traditional office hours and across multiple channels such as email and websites.
2. Conversational AI is also present in IoT devices and smart home assistants such as Amazon Echo and Apple's Siri, providing users with a seamless and convenient interface.
3. In banking, chatbots can handle complex requests, freeing up human personnel to focus on more critical tasks.
4. In the field of human resources, chatbots can automate the process of screening candidate credentials, saving time and effort for human recruiters.
5. In the healthcare industry, patients can benefit from AI-powered chatbots to describe their symptoms and conditions online, thereby reducing wait times and getting timely assistance.

‍

On the other hand, one of the significant concerns is the likelihood of bias and discrimination in the algorithms that power these chatbots. Therefore, it is important to carefully consider the potential risks and drawbacks and ensure their use is guided by ethical principles and a commitment to promoting human well-being.

One strategy to navigate the challenges arising from AI model development is to work with a trusted AI advisor. Unleashing AI is here to help. Our team of experienced AI experts can assist you in creating custom and future-proof AI solutions tailored to your specific needs while ensuring maximum return on investment. We have delivered exceptional customer service and savings, elevating conversational AI for different companies. 

Ready to incorporate AI and ML technology in your organization? Contact us today to unleash the full potential of AI for your organization.

‍