Evaluate different types of generative AI applications

Evaluate different types of generative AI applications

putilov_denis – stock.adobe.com


AI encompasses many techniques for developing software models capable of doing meaningful work, including neural networks, genetic algorithms, and reinforcement learning. Previously, only humans could do this work. Now, these techniques can create different types of AI models.

Generative AI models are one of the most important types of AI models. A generative model creates things. Any tool that uses AI to generate a new output – a new image, a new paragraph, or a new machine part design – incorporates a generative model.

The different applications of generative models

Generative AI works in a wide range of applications, including:

  • Natural language interfaces. By performing both speech and text synthesis, these artificial intelligence systems power digital assistants such as Amazon’s Alexa, Apple’s Siri and Google Assistant, as well as tools that automatically summarize text or automatically generate press releases from a set of key facts.
  • Image synthesis. These AI systems create images based on instructions or directions. If told to do so, they will create an image of a kiwi bird eating a kiwi sitting on a large padlock key. They can be used to create commercials, fashion designs, or film production storyboards. DALL-E, Midjourney and Wombo Dream are examples of AI image generators.
  • Spatial synthesis. AI can also create three-dimensional, real and digital spaces and objects. It can design buildings, rooms, and even entire city maps, as well as virtual spaces for metaverse-style play or collaboration. Spacemaker is a real-world architectural program, while Meta’s BuilderBot (in development) will focus on virtual spaces.
  • Design of products and synthesis of objects. Now that the public is more aware of 3D printing, it should be noted that generative AI can design and even create physical objects like machine parts and household items. AutoCAD and SOL75 are tools using AI to perform or assist in the design of physical objects.

Many tools leverage both generative and discriminative AI models. Discriminative models, on the other hand, identify things. Any tool that uses AI to identify, categorize, label or assess the authenticity of an artifact (physical or digital) incorporates a discriminating model. A discriminative model usually does not say categorically what something is, but rather what it is most likely based on what it sees.

Diagram of the GAN training method
GAN training method

How generative and discriminative models work together

A generative adversarial network (GAN) uses a generative model to create outputs and an adversarial discriminant model to evaluate them, with feedback loops in between. For example, a GAN may be tasked with writing fake restaurant reviews. The generative model would attempt to create seemingly real criticisms and then pass them, along with real criticisms, through the discriminative model. The discriminator acts as an adversary to the generative model, trying to identify fakes.

Feedback loops ensure that exercise trains both models to perform better. The discriminator, which is then told which entries were real and which were fake after evaluating them, adjusts to better identify fakes and not flag real reviews as fake. The generator gets better at generating undetectable fakes as it learns which fakes the discriminator successfully identified and which genuine reviews it mislabeled.

This phenomenon is applied in the following industries:

  • Finance. Artificial intelligence systems monitor transaction streams in real time and analyze them in the context of a person’s history to determine whether a transaction is genuine or fraudulent. All major banks and credit card companies now use this software; some develop their own and others use commercially available solutions.
  • Manufacturing. Factory AI systems can monitor inflows and outflows using cameras, X-rays, and more. They can report or divert parts and products that may be defective. Both Kyocera Communications and Foxconn use AI for visual inspection of their facilities.
  • Cinema and media. Just as generative tools can create fake images (e.g. a kiwi bird eating kiwi on a key), discriminative AI can identify faked images or audio files. Google’s Jigsaw division is partly focused on developing technologies to make detecting deepfakes more reliable and easier.
  • Social media and technology industry. AI systems can examine posts and post patterns to help spot fake accounts by disinformation bots or other malicious actors. Meta has been using AI for years to help find fake accounts and to flag or block pandemic-related COVID misinformation.

Generative AI may well become a widely known technology buzzword, like automation, and its myriad applications prove that this nascent branch of AI is here to stay. To meet the modern challenges facing the tech industry, it only makes sense that this technology will grow and become deeply embedded in more and more businesses.


Deepen AI business strategies




#Evaluate #types #generative #applications

Leave a Comment

Your email address will not be published.