Council: AI Agent Framework is LIVE >> Explore our GitHub

Jul 5, 2023
 min read

What is Generative AI? A Look Back — and Ahead


There's no better time to get yourself up to speed on what's out there in the growing field of generative AI — and what it could mean for you.

When we wondered if 2023 was "the year of AI" in our blog post in January, we couldn't have anticipated the rise in attention, adoption, and opportunity to unfold in five months. From rampant advancement in ‘next generation’ chatbots to growing scrutiny and conversation around ethics and regulation, it's easy to feel like a decade's worth of innovation and activity has occurred in under a year.

There's no better time to get yourself up to speed on what's out there in the growing field of generative AI — and what it could mean for you.

Modern AI: A Brief History

Chances are you've engaged with AI more than you even realize, as have others before you. Modern AI has been researched, developed, and utilized since the 1950s when early researchers and data and computer scientists like Alan Turing started measuring a machine's ability to exhibit intelligent behavior.

The next half century saw the development of rule and knowledge-based systems and programming languages, which became widely adopted in the burgeoning field of AI research. Fast forward to the decade between the 1990s and 2000s, we witnessed even more innovation in machine learning approaches, which fueled the growth of data-driven AI.

The Generative AI Journey

Throughout the last decade-plus, the industry has witnessed significant research-powered breakthroughs in AI models, where the successful application of deep machine learning approaches to tasks like computer vision and natural language processing has gradually become more refined and foundational to enabling generative AI.

During this period of development and adoption, there has been notable growth in data sets, accompanied by the implementation of more sophisticated approaches to algorithms, often involving deeper neural networks. These advancements have played a crucial role in enhancing machine learning's ability to do just that — learn — the underlying structures of data sets, enabling complex tasks such as image classification, object detection, and natural language understanding. These capabilities have become integral to the expanding list of use cases for generative AI. Furthermore, the availability of cheaper computing resources has significantly reduced the costs associated with training and running generative AI models.  

The diagram below, adapted from an article on generative AI from Sequoia Capital, visually breaks down AI into three distinct eras: development, adoption, and use cases.

Generative AI Examples — How the Future Will Look, Sound, and Support

While model approaches prior to 2015 were initially expensive and clunky to run — they already delivered impressive results across the detection and classification of unstructured data like image, text, and video. They also helped pave the way for the advanced generative capabilities we're experiencing now and powerful, user-friendly 'killer apps' like OpenAI's ChatGPT.  

Let's explore how these new models will change how AI can look, sound, feel, and support everyday needs and wants.

  • Media Creation 📹 📷 🎶

Generative AI is making it easier and more efficient than ever to create, edit, conceptualize, and generate media content across a growing number of industries for average consumers and businesses alike.

For professional designers and even just curious users looking to create fun content, text-to-image diffusion models like Stable Diffusion and Midjourney hit the mainstream to a mix of awe and even apprehension as people generated (and went viral) with self-portraits, controversial deepfakes, a variety of photo-realistic images that could fool even the sharpest of trained eyes.

In the media and advertising industry, generative models and approaches can help streamline processes, enhance creativity, and support businesses that want to generate content (copywriting, images, etc.) with a newfound sense of ease and customization. While there are concerns around the use of generative AI’s implementation in creative fields where products are sold and written — and what that could mean for creative workers and intellectual property rights — the use cases are worth paying attention to.

For video creators, generative AI models — and a growing list of consumer and business-focused applications — can synthesize videos to help users reduce editing time or create dynamic videos from text by learning the underlying pattern of video to generate new frames, graphics, captions, and more.

Across the music industry, AI’s ability to decipher sound makes it possible to remix, improvise, compose, and customize different elements of the music production process, leading to endless — and even controversial — possibilities. Just look at Grimes and imitators of Drake and The Weeknd to see what’s already hitting the AI airwaves.

  • Customer Service 🙋‍♂️

While customer service chatbots are one of the earlier conversational AI implementations, advances in generative AI can now provide even more intuitive support and options for businesses to empower customers to be more productive and creative. The development of AI tools like Microsoft 365’s Co-pilot has enabled AI models to leverage data across their family of apps to collaborate with users through conversational inputs. The space is also ripe for new entrants, startups run by veteran data scientists and engineers, building customized tools for forward thinking business leaders looking to leverage the next generation of AI chatbots. ChainML collaborated with data warehouse company, Space and Time, to leverage decentralized data and inform a conversational AI engine to meaningfully serve and support customers through powering productivity features and deliver analytics with plain text.

ChainML’s CEO Ron Bodkin recorded this demo of the Space and Time ‘Houston’ agent, powered by our Conversational AI Engine and the ChainML Protocol.

  • Self-Service Analytics 🧠

Generative AI and self-service analytics can prove useful to businesses that want to easily access and understand data. Users can describe a problem via text to automatically generate the code to query data in a database and format that information into a visualized graph to match the user's needs. In this case, generative AI and general-purpose programming languages like Python and/or data querying languages like SQL work together to turn user input into data-rich analytics. From there, as in the Space and Time example above, ChainML takes it further and builds an intuitive dashboard based on a user's analytical requirements.

  • Gaming 🎮

If you read our deep dive into generative AI in gaming, you’ll know just how much potential there is for AI to create context, lore, soundtracks, and content from the ground up. Similar to how AI can be used to support music producers, film editors, designers, and photographers, generative AI can work alongside game developers and designers to cut time and costs while unlocking vast creative opportunities.

The Path Ahead for Generative AI (and You?)

And that’s just the tip of the generative AI iceberg. While the opportunity to use generative AI models to solve problems is growing, innovation comes with challenges. Whether you’re an enterprise decision maker, developer, or everyday consumer, more work is required to address these challenges in meaningful ways. From evolving regulations, privacy considerations, new operating models, to the intricacies around performance, the only constant is change. But we’ll leave that for another article. Stay tuned.  

Stay connected to ChainML on Twitter and on our website.