Category: Article

What is Conversational AI and how did we get here?

Conversational AI is a branch of artificial intelligence that simulates human conversations.

In order to understand what conversational AI is, it is essential to know the concept of Natural Language Processing and how it has evolved to allow us to talk about conversational AI.

Natural Language Processing

Natural language processing seeks to teach machines to understand, interpret and generate human language responses. Its beginnings date back to the 1950s, when researchers started working with rule-based approaches and formal grammars to analyze and generate text. However, these systems were very limited as they relied on predefined rules and could not adapt to variations or ambiguities of natural language.

In order to solve these problems, important advances have been made in recent years, of which we can highlight some key moments.

Neural Networks

In the mid-1980s, artificial neural networks applied to NLP began to be used. These were inspired by the structure and functioning of the human brain to develop models capable of learning complex patterns and structures from data. As the availability of large amounts of digitized text data and computational capacity increased, machine learning became an essential tool for the evolution of NLP.

Deep Learning

Already at the beginning of the 21st century, NLP underwent a significant transformation thanks to the application of deep algorithms, what we know as Deep Learning. These models improved the ability of systems to understand and generate natural language text, as well as making them more robust and accurate, which is why they began to be applied to tasks such as machine translation, text summarization, text generation or document classification.

Transformers

The big change came with the introduction of Transformers in 2017. The most remarkable thing about this architecture is that it contains an attention matrix that evaluates the relationship of each word with all the other words in the sentence, thus taking into account the whole context and allowing to grasp the meaning of the text as a whole.

Generative Models

And I’m sure by now you’re wondering at what point we’re going to talk about the famous ChatGPT. Well, the emergence of generative language models such as ChatGPT or LLaMa, introduced in late 2022 and early 2023, are based on an architecture containing several layers of Transformers, which implies that they have hundreds of millions of parameters and also require a corpus with billions of words for training. These models have marked a before and after in AI and are generating strong discussions regarding their impact, as their capabilities in terms of text generation are almost comparable to those of a human.

Once we have traced the evolution of natural language processing, it is easier to understand how we have come to develop what we know today as conversational AI. This is based on models that allow the understanding of language and context, as well as the generation of responses according to the conversation and similar to those that a human would give.

Would you like to know what a Conversational AI can do for you? We invite you to meet MAIA.

26/10/2023
Generative Artificial Intelligence
What is GenAI and who are the main players?

For several years now, Artificial Intelligence has been revolutionising our society and transforming the way we interact with technology. But many people call things “Artificial Intelligence” that are not Artificial Intelligence… so what is Artificial Intelligence?

The concept of Artificial Intelligence was born in the 1950s, when pioneers such as Alan Turing and John McCarthy began to explore the possibility of creating machines that could simulate human thought. It was at that time when it was determined that AI is the ability of machines to learn, to reason and to make decisions autonomously, imitating human intelligence. Over the last few years, Artificial Intelligence has grown spectacularly and we see it in a lot of things and the concept of Generative Artificial Intelligence has become a trending topic.

Generative Artificial Intelligence

Generative AI is a branch of artificial intelligence that focuses on creating original and autonomous content, such as images, music and text, from input data. The difference between traditional AI and generative AI is that the first is based on rules and historical data and the second uses neural networks to produce new results.

Generative AI has been in development for several years, but it was the rise of ChatGPT that made it popular. The influence of Open AI’s GPT-3 (Generative Pre-trained Transformer 3) language model has been abysmal, setting an important milestone for the high quality of the texts generated thanks to prompts.

But what is a Prompt? A prompt is a question, a request, an instruction to an AI program to generate a result. Depending on the AI you use, you can create articles, images, music, videos or programming code.

Principal players

A number of companies are currently leading the way in the development of generative AI. These include:
- Google: has developed a number of generative AI models, such as BARD based on the LaMDA experimental language model designed by Google specifically for dialogue applications. Like ChatGPT, it is a conversational system where the user interacts with the model through regular messages and receives an output.
- OpenAI: Well known for its ChatGPT, OpenAI has also developed DALL-E 2, which is an interface that is able to generate realistic images from textual descriptions.
- Meta: Mark Zuckerberg’s company also has its own Generative AI product called Llama 2, which it has recently incorporated into its company products.
- Adobe: has developed the Sensei model, capable of generating creative content, such as videos and animations, and Firefly, which, from just a few words, can create graphics, use infinite colour combinations, fill or enlarge images and generate text effects.
- Microsoft: The Redmond-based company has Github Copilot, which is an AI assistant that helps develop code. It also has the Microsoft 365 Copilot product, which is the office suite assistant that makes certain types of tasks easier.
The truth is that every day more and more companies, using AI, help us to make our lives easier and increase our productivity. Like MAIA, which is a conversational AI capable of optimising the customer experience, improving customer satisfaction by managing complex processes in shorter times and in any contact channel.

MAIA is the conjunction of technology and human experience; it provides, first-hand, what users need thanks to the data and the interpretation of the people who work with them on a daily basis. Do you want to know more about MAIA? You can do it here.
20/10/2023
Chatbots vs. Conversational Artificial Intelligence

In our busy day, we often have to contact our bank, cell phone provider, or even the gym! In addition, we are often short of time, and therefore want to resolve our queries and formalities as quickly as possible.

Before making a phone call, we make use (voluntarily or involuntarily) of chat or messaging applications such as WhatsApp. In this way, companies provide us with the information we need without having to talk or interact with a person, and for this they use “chatbots”.

“Chatbot” is a very common term to name, in a standard way, all the technologies that allow to automate the answers in a chat without the need for human intervention.

So, when we present MAIA as a conversational Artificial Intelligence, it is normal that we encounter the following questions from some customers:

Is a Chatbot the same as a Conversational AI? What are the differences between one and the other?

Let’s start with the most important: No, a Chatbot and a Conversational Artificial Intelligence are not the same, but what is the difference?

Conversational Artificial Intelligences are mainly characterized by capabilities such as:

Speech recognition.

Unlike chatbots, conversational AI’s are designed so that, in addition to processing written language, they can identify spoken language.

Chatbots are usually limited to written channels such as chat or messaging services (WhatsApp, Telegram, etc).

Natural Language Understanding (NLU).

A Conversational Artificial Intelligence, it is able to recognize a customer’s requests when he expresses himself in his own words, so it is open to recognize complex requests. Not everyone says the same thing in the same way. For example.

In the case of chatbots, there is no comprehension capability, the way a customer interacts is through predefined options or a decision tree that guides the conversation through specific topics.

Dialogue management.

Conversational AI’s are dialog-oriented. They manage interactions dynamically. They also have the ability to “remember” what they have discussed in previous requests with a customer, pick up on topics, because they manage the context of the entire conversation.

In contrast, a chatbot manages interactions in a fixed, rule-based way, often guided by menus of options, with no flexibility or ability to go outside the flow. Chatbots, moreover, do not have the ability to remember previous queries to execute or respond to complex requests.

Natural Language Generation (NLG).

A conversational AI responds using human and natural language to customer requests and can adapt to the way of communicating in a given situation. While a chatbot gives predefined answers that do not vary or adapt to the context that the customer may have provided.

Machine Learning (ML).

The use of Machine Learning allows conversational AI to constantly learn from responses, conversation history and user preferences; this allows to personalize responses and user experience. It also allows the quality of responses to constantly evolve because AI learning is parallel to its operation.

Technicians take advantage of all the experience acquired by Conversational AI to teach it what to say in new situations not previously identified.

Chatbots do not learn, their responses are static so they do not allow customizing the user experience; they also need manual programming to keep them updated.

The future… Conversational Artificial Intelligence or Chatbots?

What is certain is that both Chatbots and Conversational AI’s are here to stay.

Both allow companies to automate information processes in marketing, sales and after-sales services, mainly thanks to their ability to provide 24/7 attention.

However, with the recent emergence of Open AI and Chat GPT, and the visibility that is being given to the capabilities provided by Generative AI models, many companies have considered using, or evolving their chatbots, and implementing Conversational AI.

But what is the main reason? It is undoubtedly to provide a better customer experience, more personalized attention and, above all, to attend to more complex processes in an automated way, resulting in significant savings.

Do you have any doubts about why you should choose a Conversational AI?

If so, we’d be happy to talk to you! Shall we talk?

11/10/2023