5 minutes to grasp the key to the explosion of generative AI in 2023: from machine learning to ChatGPT

Mark Weiser, a pioneer in computer science, once said, "The most profound technologies are those that disappear and weave themselves into the fabric of everyday life." With the optimization of ChatGPT, the power of generative AI has truly come to the forefront this year! In comparison to many technologies that gradually integrate into everyday life and become invisible, when NVIDIA CEO Jensen Huang declares, "The era of generative AI has arrived!" The next great technology has officially arrived!

Since the launch of ChatGPT 3.5 by OpenAI in late 2022, the power of generative AI has been on the rise. In February 2023, Microsoft introduced Bing, a generative AI that can be used on mobile phones. Then, in the spring, GPT-4 was unveiled, greatly boosting the usage of ChatGPT. The potential of generative AI has fully blossomed this year. Have you, too, tried using it and been amazed by the intelligent responses and speed of the new chatbots?

Chatbots have long been a part of our daily lives. When you open a messaging app like LINE and inquire about new product information from an official brand account late at night, you may receive an automatic reply from a chatbot. But why aren't you impressed by the conversation in between? This brings us back to machine learning (ML) and generative artificial intelligence (AI).

In the following 5 minutes, we will guide you through the key points of the generative AI explosion in 2023.

Until the end of last year, it was generally believed that machine learning and artificial intelligence were only relevant to technology enthusiasts, creative professionals, large corporations, or data-driven companies. However, in just a few months of word-of-mouth accumulation, as we entered the early part of this year, an increasing number of people began using ChatGPT, causing a surge in generative AI.

Users of ChatGPT span various demographics, including students, working professionals, managers, and programmers, cutting across age, gender, and industries. The number of ChatGPT users is comparable to that of many trending products. Whether it's spontaneously asking ChatGPT to translate a short text or having it search for extensive data, ChatGPT's capabilities are vast.

Of course, regardless of the accuracy of the content generated by ChatGPT, before discussing the content it produces, we will first delve into why this is happening in this year. Why did humans suddenly dive into the whirlwind of generative AI? Beyond ChatGPT, other forms of generative artificial intelligence have also started emerging, and they are not limited to text translation and content generation. The range of applications has expanded to include images, audio, artistic design, and various forms of data.

Sam Altman, the entrepreneur who gained notoriety for co-founding OpenAI, is frequently interviewed. In a recent public speech in Taiwan, he discussed the generative AI revolution brought about by OpenAI, likening it to the launch of the first-generation iPhone, which transformed the smartphone market and even spurred the growth of apps and new service models. He believes that artificial intelligence technology will similarly lead to new market dynamics

In fact, generative artificial intelligence and machine learning are both AI applications. Generative artificial intelligence is a framework within machine learning, but not all machine learning frameworks are generative artificial intelligence. Before praising ChatGPT for its role as a virtual assistant, it may be worthwhile to first understand its history and relationship with AI.

Google has defined the differences among these three:

Artificial Intelligence (AI): This is the most general concept that enables machines or systems to perceive, reason, act, or adapt like humans.

Machine Learning (ML): It is an AI application that uses algorithms to analyze data, thereby extracting knowledge from data and learning automatically.

Generative AI: It refers to the use of AI (Artificial Intelligence) to create new content, such as text, images, music, audio files, and video files.

Generative AI is a form of artificial intelligence that, according to IBM's definition, refers to deep learning models capable of generating high-quality text, images, and other content based on trained data.

Deep Learning is a subset of machine learning (ML) that uses specific algorithmic structures inspired by the human brain, known as neural networks. Deep learning has driven many advancements and applications in artificial intelligence.

Reflecting on the progress of AI through the lens of Generative AI, we can understand that AI is continually advancing rapidly. Generative artificial intelligence represents a significant milestone in the field of artificial intelligence. Looking back at the history of AI development, the rise of generative AI signifies that AI is moving towards broader and more complex real-world applications. However, the advantages and disadvantages of its application are issues that cannot be ignored.Currently, there is still a gap in achieving full automation and complete accuracy in AI. Generative AI has opened up exciting possibilities, but there are challenges to overcome.

Here is a brief introduction to the top 10 popular generative AI tools according to Technology magazine's ranking. These top generative AI tools empower businesses and professionals. However, it's important to note that some of these tools may have regional restrictions limited to Europe or the Americas. Additionally, many of these tools are primarily offered by the three major enterprises: OpenAI, Google, and Microsoft. It can be said that 2023 is also a battleground for generative AI products.

1.ChatGPT (OpenAI): Known for its conversational generative content, ChatGPT has become the fastest-growing consumer software application in history, with over 100 million users as of January 2023.

2.Bard (Google): Developed by Google as a generative AI chatbot, Bard was created as a competitor to OpenAI's ChatGPT and expanded to more countries by May 2023.

3.Claude 2 (Anthropic): Developed by Anthropic, Claude 2 is considered one of ChatGPT's main competitors. It can assist with content summarization, document handling, and answering questions. Claude 2's data has been updated to early 2023.

4.GitHub Copilot (OpenAI & Microsoft): Powered by OpenAI's technology, Copilot is an automatic code completion service launched by GitHub in the past year, assisting developers with code writing using AI.

5.Bing AI (Microsoft): Microsoft's Bing AI allows users to chat, search, and find inspiration. It includes features such as providing summary answers, drafting content, and viewing images.

6.Sensei GenAI (Adobe): Developed by Adobe's Sensei GenAI, it can directly integrate new creative generative AI models like Firefly Adobe Sensei. It supports content personalization, editing, marketing copy generation, and interactive experiences.

7.DALL-E 2 (OpenAI): Also from OpenAI, DALL-E 2 is an AI drawing software developed from the GPT-3 model. It can create realistic photos or artwork based on natural language descriptions.

8.AlphaCode (Google DeepMind): AlphaCode from Google DeepMind can autonomously write code.

9.Autodesk Generative Design: Used across various industries from automotive to architecture, Autodesk's Generative Design allows users to quickly generate optimized high-performance design alternatives.

10.Synthesia: It supports text-to-3D technology development and can generate a large number of realistic images.

Despite the remarkable capabilities of generative AI in self-learning and data processing, it still requires human guidance, training, and supervision to ensure its effectiveness and ethical application. Humans play a crucial guiding role in the optimization process of generative AI.

In 1950, Alan Turing, the father of computer science, posed the question "Can Machines Think?" Fast forward 73 years, and we now have chatbots capable of simulating human brains, engaging in conversational generative content, and more. However, chatbots operate in various ways, depending on the complexity of their software.

Next, let's delve deeper into the current champion of this wave of generative AI battle - ChatGPT. It is a type of chatbot, essentially an evolved version.

A standard chatbot is a computer program designed to engage in conversations or text-based interactions with users. Early classic chatbots include "ELIZA" from 1966 and "PARRY" from 1972. In recent years, examples include Apple's "Siri" and Amazon's "Alexa." Chatbots are capable of simulating human conversations and serve practical purposes such as customer service and information retrieval. Common applications of chatbots in the market include CRM personalization, real-time customer support, shopping assistance, and payment processing, among others. In recent years, chatbots have expanded their applications into various fields, including healthcare, finance, education, and many others.

Some chatbots are equipped with natural language processing systems, while simpler systems extract keywords from user inputs and find the most appropriate response from a database. Currently, chatbots are often considered virtual assistants.

ChatGPT, short for "Chat Generative Pre-trained Transformer," differs from standard chatbots that rely on pre-programmed responses. It leverages extensive training to provide contextually meaningful and logically sound responses. It can understand language elegantly, grasp the intent behind user queries, and provide relevant replies.

"ChatGPT can complement human expertise by providing real-time access to knowledge, recommending solutions, and automating repetitive tasks." Since the data training for ChatGPT3.5 concluded before September 2021, data collection is based on information available up to that point. However, there are now plugins that can overcome this limitation. Nonetheless, in comparisons between GPT-4 and ChatGPT3.5, it is generally believed that GPT-4 is more reliable, creative, and capable of handling more nuanced instructions, especially for complex tasks.

Will AI replace human jobs? Let's first see how much time it can save us! The explosive growth of generative AI is closely linked to its collaboration with humans. According to Salesforce's survey, 52% of people say they now use generative AI more frequently than when they first started. "Using generative AI, on average, saves 5 hours per week (52 weeks in a year) / 8 hours per workday = an average of 32.5 days or more than a month saved each year."

Indeed, as Steve Jobs devoted his life to the idea that "technology should change the lives of people" and Mark Weiser's belief that "the greatest technology is the one that has integrated into everyday life and gradually disappeared."

Great technology should be widely applied in people's lives. As generative AI matures on the software side, it becomes a "virtual assistant" that anyone can use, greatly changing our lifestyle and creative applications. When major companies like OpenAI, Google, Microsoft, and others are competing fiercely in the development and optimization of generative AI, you and I as users are beneficiaries. Most programmers believe that "using GPT is an efficiency issue. Of course, it has its flaws, but it excels in handling conversations, which is why it has been so successful this year. Otherwise, we would just use a regular ChatBot." Instead of worrying about AI replacing your job, it's better to think about how to make good use of technology assistance and collaborate effectively with virtual assistants in "smart automation" roles.