A quick scan of the headlines makes it seem like generative artificial intelligence is everywhere these days. In factsome of those headlines may actually have been written by generative AIlike OpenAI’s ChatGPTa chatbot that has demonstrated an uncanny ability to produce text that seems to have been written by a human.
But what do people really mean when they say “generative AI?”
Before the generative AI boom of the past few yearswhen people talked about AItypically they were talking about machine-learning models that can learn to make a prediction based on data. For instancesuch models are trainedusing millions of examplesto predict whether a certain X-ray shows signs of a tumor or if a particular borrower is likely to default on a loan.
Generative AI can be thought of as a machine-learning model that is trained to create new datarather than making a prediction about a specific dataset. A generative AI system is one that learns to generate more objects that look like the data it was trained on.
“When it comes to the actual machinery underlying generative AI and other types of AIthe distinctions can be a little bit blurry. Oftentimesthe same algorithms can be used for both,” says Phillip Isolaan associate professor of electrical engineering and computer science at MITand a member of the Computer Science and Artificial Intelligence Laboratory (CSAIL).
And despite the hype that came with the release of ChatGPT and its counterpartsthe technology itself isn’t brand new. These powerful machine-learning models draw on research and computational advances that go back more than 50 years.
An increase in complexity
An early example of generative AI is a much simpler model known as a Markov chain. The technique is named for Andrey Markova Russian mathematician who in 1906 introduced this statistical method to model the behavior of random processes. In machine learningMarkov models have long been used for next-word prediction taskslike the autocomplete function in an email program.
In text predictiona Markov model generates the next word in a sentence by looking at the previous word or a few previous words. But because these simple models can only look back that farthey aren’t good at generating plausible textsays Tommi Jaakkolathe Thomas Siebel Professor of Electrical Engineering and Computer Science at MITwho is also a member of CSAIL and the Institute for DataSystemsand Society (IDSS).
“We were generating things way before the last decadebut the major distinction here is in terms of the complexity of objects we can generate and the scale at which we can train these models,” he explains.
Just a few years agoresearchers tended to focus on finding a machine-learning algorithm that makes the best use of a specific dataset. But that focus has shifted a bitand many researchers are now using larger datasetsperhaps with hundreds of millions or even billions of data pointsto train models that can achieve impressive results.
The base models underlying ChatGPT and similar systems work in much the same way as a Markov model. But one big difference is that ChatGPT is far larger and more complexwith billions of parameters. And it has been trained on an enormous amount of data — in this casemuch of the publicly available text on the internet.
In this huge corpus of textwords and sentences appear in sequences with certain dependencies. This recurrence helps the model understand how to cut text into statistical chunks that have some predictability. It learns the patterns of these blocks of text and uses this knowledge to propose what might come next.
More powerful architectures
While bigger datasets are one catalyst that led to the generative AI booma variety of major research advances also led to more complex deep-learning architectures.
In 2014a machine-learning architecture known as a generative adversarial network (GAN) was proposed by researchers at the University of Montreal. GANs use two models that work in tandem: One learns to generate a target output (like an image) and the other learns to discriminate true data from the generator’s output. The generator tries to fool the discriminatorand in the process learns to make more realistic outputs. The image generator StyleGAN is based on these types of models.
Diffusion models were introduced a year later by researchers at Stanford University and the University of California at Berkeley. By iteratively refining their outputthese models learn to generate new data samples that resemble samples in a training datasetand have been used to create realistic-looking images. A diffusion model is at the heart of the text-to-image generation system Stable Diffusion.
In 2017researchers at Google introduced the transformer architecturewhich has been used to develop large language modelslike those that power ChatGPT. In natural language processinga transformer encodes each word in a corpus of text as a token and then generates an attention mapwhich captures each token’s relationships with all other tokens. This attention map helps the transformer understand context when it generates new text.
These are only a few of many approaches that can be used for generative AI.
A range of applications
What all of these approaches have in common is that they convert inputs into a set of tokenswhich are numerical representations of chunks of data. As long as your data can be converted into this standardtoken formatthen in theoryyou could apply these methods to generate new data that look similar.
“Your mileage might varydepending on how noisy your data are and how difficult the signal is to extractbut it is really getting closer to the way a general-purpose CPU can take in any kind of data and start processing it in a unified way,” Isola says.
This opens up a huge array of applications for generative AI.
For instanceIsola’s group is using generative AI to create synthetic image data that could be used to train another intelligent systemsuch as by teaching a computer vision model how to recognize objects.
Jaakkola’s group is using generative AI to design novel protein structures or valid crystal structures that specify new materials. The same way a generative model learns the dependencies of languageif it’s shown crystal structures insteadit can learn the relationships that make structures stable and realizablehe explains.
But while generative models can achieve incredible resultsthey aren’t the best choice for all types of data. For tasks that involve making predictions on structured datalike the tabular data in a spreadsheetgenerative AI models tend to be outperformed by traditional machine-learning methodssays Devavrat Shahthe Andrew and Erna Viterbi Professor in Electrical Engineering and Computer Science at MIT and a member of IDSS and of the Laboratory for Information and Decision Systems.
“The highest value they havein my mindis to become this terrific interface to machines that are human friendly. Previouslyhumans had to talk to machines in the language of machines to make things happen. Nowthis interface has figured out how to talk to both humans and machines,” says Shah.
Raising red flags
Generative AI chatbots are now being used in call centers to field questions from human customersbut this application underscores one potential red flag of implementing these models — worker displacement.
In additiongenerative AI can inherit and proliferate biases that exist in training dataor amplify hate speech and false statements. The models have the capacity to plagiarizeand can generate content that looks like it was produced by a specific human creatorraising potential copyright issues.
On the other sideShah proposes that generative AI could empower artistswho could use generative tools to help them make creative content they might not otherwise have the means to produce.
In the futurehe sees generative AI changing the economics in many disciplines.
One promising future direction Isola sees for generative AI is its use for fabrication. Instead of having a model make an image of a chairperhaps it could generate a plan for a chair that could be produced.
He also sees future uses for generative AI systems in developing more generally intelligent AI agents.
“There are differences in how these models work and how we think the human brain worksbut I think there are also similarities. We have the ability to think and dream in our headsto come up with interesting ideas or plansand I think generative AI is one of the tools that will empower agents to do thatas well,” Isola says.
