Is Generative AI the Unicorn of AI Technologies?

The recent surge in interest in generative artificial intelligence (AI), through its most exquisite expression so far, ChatGPT, has stirred debate on AI around the world. We’ve been witnesses of high levels of excitement as well as anxiety related to the amazing capabilities of a large language model to respond to whatever it’s been asked.

Critics far and wide have taken the opportunity to reiterate that AI, including large learning models, will continue to perpetuate societal biases that are present in the data used to train them and that it can be easily used for malicious purposes, like spreading disinformation and misinformation, or invading people's privacy.

On the other hand, AI aficionados are celebrating the emergence of a new piece of technology that stands out from the crowd and will forever change the way creativity is perceived. ChatGTP has even raised questions about whether creativity is the fundamental prerogative of humankind or not?

But before jumping into the pros and cons, let’s go through the terminology.

What is Generative AI?

As the name suggests, generative AI is a branch of artificial intelligence that focuses on producing new pieces of content such as text, images, music, speech, code, or video. It can be either supervised or unsupervised; supervised learning requires labelled data while unsupervised learning does not. With supervised learning, the input data must have labels associated with it, allowing the program to learn specific relationships among different objects in the data. Unsupervised learning relies on continuous analysis of data points without labels, searching for hidden patterns within them.

How Does Generative AI Work?

Generative AI systems are composed of multiple elements – out of which the most important are a generator (or model) and a discriminator (or critic). The generator is responsible for creating novel content by taking in random input; this input then goes through several layers until an output is produced. The discriminator evaluates this output against given criteria to determine its accuracy and quality. If the output meets a certain threshold for quality, then it is considered valid and will be used in further training iterations.

The process of machine learning happens when both components are trained simultaneously through optimization algorithms to improve performance over time, as both components refine their strategies for producing more accurate results. Becoming increasingly better at generating new content within a given domain, they also become increasingly better at discriminating between generated samples and real-world examples; this allows them to accurately recognize new inputs and generate outputs that are more accurate than before.

Machine-learning techniques behind generative AI have evolved over the past decade; deep learning and General Adversarial Network (GAN) have been typically used, while the latest and most successful approach are generative pretrained transformers (GPTs).

A?Generative Pretrained Transformer?(GPT) is a type of large language model (LLM) that uses deep learning to generate human-like text. They are called ‘generative’ because they can generate new text based on the input they receive, ‘pre-trained’ because they are trained on a large corpus of text data before being fine-tuned for specific tasks, and ‘transformers’ because they use a?transformer-based neural network architecture to process input text and generate output text.

What is a Large Language Model?

Large learning models (also known as deep learning models) are types of artificial neural networks which consist of multiple hidden layers of neurons. This gives them the incredible capability to learn complex relationships between data points by utilizing all the layers and connections within them, allowing them to accurately capture patterns and generate predictions more accurately than simpler methods.

These artificial intelligence tools can?read, summarize, and translate text that predict future words?in a sentence, allowing the tech to generate sentences eerily similar to how humans write and talk.

ChatGPT Is the Absolute Expression of Generative AI so Far

No wonder ChatGPT created so much buzz. ChatGPT is developed on a?large language model (LLM) ?and is powered by the most advanced transformer-based neural network architecture. It has been designed to generate human-like conversations as well as creative and engaging content such as short stories, articles, tweets, and blog posts. And, yes, it can answer pretty much everything it is asked.

In addition, ChatGPT is able to learn from the conversation it has been having with a user, allowing it to continue learning and improving its prediction accuracy over time. This unique capability allows it to provide better predictive results for longer conversations without having to be re-trained or reset — making it ideal for applications like conversational agents or interactive fiction writing. Although ChatGPT is not connected to the internet, it is trained on an enormous amount of historical data (pre-2021) and presents its findings in an easily consumable way.

Applications of Generative AI

Generative AI models are incredibly diverse. They can take in various content such as images, longer text formats, emails, social media content, voice recordings, program code, and structured data. They can output new content, translations, answers to questions, sentiment analysis, summaries, and even videos… you name it! These universal content machines have many potential applications in business, and today marketing applications are among the most common uses of generative AI. In the future, there is potential for generative AI to make an impact in healthcare and life sciences—to make diagnoses, for example, or find new cures for disease. Here are some use cases.

Marketing Applications

These generative models are most used, at least for the time being, for marketing applications. Jasper, for example, a marketing-focused version of GPT-3, can produce blogs, social media posts, web copy, sales emails, ads, and other types of customer-facing content. Customers say that Jasper helped them ramp up their content strategy , and allowed writers more time to do better research, ideation, and strategy.

Moreover, text-to-image programs such as DALL-E, Midjourney, or Stable Diffusion are forever changing how art, animation, gaming, movies, and architecture among others are being rendered. Art directors from the advertising world are already using it to create product/brand images that sell better. For example, Nestle used an AI-enhanced version of a Vermeer painting to help sell one of its yogurt brands.

Code Generation Applications

GPT-3 has also proven to be an effective, if not perfect, generator of computer program code. For example, Microsoft-owned GitHub Copilot which is based on?OpenAI’s Codex model , suggests code and assists developers in autocompleting their programming tasks. The system has been quoted as autocompleting up to 40% of developers’ code , considerably augmenting the workflow.

Conversational AI Applications

LLMs are increasingly being used at the core of conversational AI or chatbots. They can improve the levels of understanding of conversation and context awareness and lead to better business problem solving. Facebook’s?BlenderBot , for example, which was designed for dialogue, can carry on long conversations with humans while maintaining context. Google’s BERT ?is used to understand search queries and is also a component of the company’s DialogFlow chatbot engine.

Knowledge Management Applications

One emerging application of generative AI is to be employed to manage text-based (or potentially image or video-based) knowledge within an organization. The labor intensiveness involved in creating structured knowledge bases has made large-scale knowledge management difficult for many large companies. However,?some research ?has suggested that LLMs can be effective at managing an organization’s knowledge when model training is fine-tuned on a specific body of text-based knowledge within the organization. The knowledge within an LLM could be accessed by questions issued as prompts. Morgan Stanley, for example, is working with OpenAI’s GPT-3 to fine-tune training on wealth management content , so that financial advisors can both search for existing knowledge within the firm and create tailored content for clients easily.

Industry-Specific Use Cases in the Near Future?

Over the next several years, we will see generative AI permeating traditional industries, like manufacturing, health care, and pharmaceuticals, for example. Once a generative model has been trained, it can be fine-tuned for specific content domains with much less data. We are now starting to see specialized generative models for biomedical content, legal documents, and translated text, which will give rise to additional use cases in those industries and domains.

Nevertheless, Using Generative AI Isn’t Without Risks

Generative AI systems can rapidly lead to a number of legal and ethical issues. “Deepfakes,” or images and videos that are created by AI and purport to be realistic but are not, have already arisen in media, entertainment, and politics.?

Additionally, generative AI raises questions about what original and proprietary content is and may have a significant impact on content ownership. For example, digital artist Greg Rutkowski fears that the internet will be flooded with artwork that is?indistinguishable from his own , simply by telling the system to reproduce an artwork in his unique style. Professor of art Carson Grubaugh shares this concern and predicts that large parts of the creative workforce, including commercial artists working in entertainment, video games, advertising, and publishing, could?lose their jobs because of generative AI models .

Generating false and misleading content. This is even more alarming when seen in the context of automated troll bots, with capabilities advanced enough to render obsolete the Turing test? – which tests a machine’s ability to exhibit intelligent behavior similar to or indistinguishable from a human. Such capabilities can be misused to generate?fake news and disinformation across platforms and ecosystems .

Below is ChatGPT’s confession about its own limitations:

To Conclude

Despite layoffs in the technology sector, generative AI companies continue to receive interest from investors.?Sequoia thinks that?the field of generative AI can generate trillions of dollars in economic value , making it sound as if generative AI is the new unicorn of AI technologies. No less than 150 start-ups?have emerged and are already operating in the space.

Large models continue to be trained on massive datasets represented in books, articles and websites that may be biased in ways that can be hard to filter completely. Despite substantial reductions in harmful and untruthful outputs achieved by the use of?reinforcement learning from human feedback?(RLHF) in the case of ChatGPT, OpenAI acknowledges that their models can still generate biased outputs.

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

?? Hi, we are IRPAAI,

??We are a neutral industry professional association focused on providing education, networking and advice to executives on RPA, intelligent automation & enterprise AI. We are a community of 130,000 members worldwide.

??We offer leading-edge market intelligence, coaching and access to resources, allowing business leaders to position themselves at the cutting edge of knowledge work and technology.

Here are some of our others posts we think you will enjoy too:

?? Intelligent Automation: Your Roadmap to Successful Transformation -?https://www.dhirubhai.net/pulse/intelligent-automation-your-roadmap-successful-daniel-goodstein

?? Robotic Process Automation: Lessons from the Pandemic -?https://www.dhirubhai.net/pulse/robotic-process-automation-lessons-from-pandemic-daniel-goodstein

??? The Digital Enterprise Institute: The Digital Leader Show: AI Business Use Cases with Twilio’s Shreshth Sharma-?https://lnkd.in/dVQcrWYv

End.