Generative AI

What is generative AI?

Generative AI is a type of artificial intelligence (AI) that uses machine learning (ML) and deep learning algorithms to produce new content such as text, images, video, music, and computer code in response to users’ prompts and questions. It has the potential to dramatically transform the way humans create original content for personal, business, and artistic purposes.

Generative AI is different from retrieval augmented generation (RAG) AI in that it is a subset of RAG. RAG combines the strengths of both generative AI and retrieval AI. Generative AI is also different from cognitive AI, which mimics the way the human brain works to get its results.

How does generative AI work?

Before it can create new content, a generative AI system must first be trained. Vast quantities of information—in the form of words, images, music, or other content—are fed into a deep learning system.

From there, an AI neural network, which is a subset of ML that teaches computers to process data by mimicking the human brain, sifts through the data, helping the system “learn.” For example, the popular ChatGPT chatbot was trained on massive datasets and more than 300 billion words gathered from the internet, books, magazines, songs, plays, movies, and other publicly available data sources. ChatGPT studies and learns from all this data to discern patterns and structures.

Once the generative AI model has assimilated enough knowledge, it can begin creating. Based on the patterns it extrapolates from the data it was trained on, it can now generate new content based on directives or questions from users. For example, if it was trained on the novels of a particular author, it could write text in the style and use the subject matter found in that author’s novels if prompted to do so by a user. If someone wants a new image, they can enter some characteristics and generative AI can produce a drawing.

Because it is trained to be continuously learning, generative AI produces multiple versions of content and chooses the best one to proceed with at the time. This process of refinement is what makes generative AI systems capable of constantly improving the quality of their output.

What are the benefits of generative AI?

There are many potential benefits of generative AI, including the following:

• Accelerate content creation — Whether doing graphic design, producing a training video, writing an article, composing a song, or debugging code, content that can be rapidly generated with appropriate user inputs— questions, instructions, visual images, or notes—is fast becoming one of the most popular benefits of using generative AI. Gartner predicts that 30% of outbound marketing messages will be created by generative AI by 2025, up from just 2% in 2022.
• Enhance efficiency of human workers — Even when the initial output of a generative AI model is not up to quality standards, humans can take the raw content and polish it to their satisfaction in a much shorter time than it would take to do from scratch.
• Increase organizational productivity — Because generative AI models can work 24/7, never tire, never get sick, and never need breaks, more work can get done around the clock, boosting overall productivity.
• Cut costs — Although some companies may use generative AI to replace job functions and therefore reduce expenses, many others may use generative AI to assist their existing workers to achieve more than was previously possible.
• Improve the customer experience — By providing warmer, more personalized, and human-like responses to customer questions and support issues, generative AI customer service chatbots can enhance the customer journey.
• Accelerate new product development — Generative AI can speed up the development and discovery of new drugs, simplify enterprise application coding, empower innovation, and jumpstart new product ideas.
• Enhance training and education — With customized testing and dynamic assignments of training and learning materials, generative AI can personalize educational experiences for both students and employees in a way that supports faster and more effective learning.
• Support better decision-making — Because of generative AI’s capability to ingest large amounts of data and spot patterns and trends, it can help human executives gain insights that they otherwise might have missed, helping them make better decisions or formulate more effective strategies.

What are some risks and limitations of generative AI?

Although generative AI has numerous benefits, it also comes with certain risks and limitations if not managed appropriately and used responsibly. The following are some of them:

• Sub-standard quality — While generative AI can deliver content rapidly, the quality may not always meet required standards. It requires meticulous fact-checking, monitoring, and editing to ensure the output is accurate, relevant, and of high-enough quality.
• Uniformity and conformity rather than creativity — AI can mimic the patterns it has learned from ingesting a broad range of content, but it doesn’t really understand what it has done. Human creativity, and especially intuition and the ability to make leaps of logic, will limit the depth and emotional resonance of what generative AI produces.
• Unintentional revealing of private, proprietary, or sensitive information — Generative AI needs vast amounts of data for training its models, so businesses and individuals have to be very cognizant of the data sources used to train those models as to not inadvertently expose private or personally identifiable information (PII), leak sensitive information, or violate data privacy and security regulations.
• Ethical misuse — There’s also the risk that AI can be used for malicious or criminal purposes, such as creating “deep fakes” (fake videos that appear real), distributing fake news, or plagiarizing content protected by copyright.
• Biased, flawed, or downright false results — As has been demonstrated in early AI models, the content that generative AI produces depends heavily on the data it is trained on to use for its output. If that data is of bad quality, exclusionary, biased, or just plain wrong, then the results that the AI system responds with will be the same.
• Social disruption — As with any form of automation, some speculate that generative AI could replace more jobs and potentially affect a much broader group of professions than previous generations of automated technologies have. History has shown that these types of industrial revolutions have great potential to create social divides and instantiate tension between older and newer generations.
• Uncertain legal terrain — There can be legal issues around the content generated by AI, such as copyright infringement or liability for publishing proprietary, harmful, or misleading content.

For all these reasons, businesses will need to be careful to use generative AI responsibly and ethically.

How is generative AI used?

Generative AI is already being used across many professions and industries:

• Creative Services / Marketing — Businesses and agencies alike are using generative AI to develop targeted advertisements, write marketing content, plan campaigns, measure campaign performance, and create websites.
• Retail — For some time, retail has been using AI to give customers purchasing recommendations based on their surfing or buying behaviors. Generative AI can also predict which products will be in high demand, at what time, and where to optimize inventory in warehouses and stores.
• Healthcare — Generative AI tools can help doctors offer personalized patient care based on a patient’s health history, including lab test results and images such as x-rays or CT scans. These tools can also help doctors detect diseases earlier due to the vast amounts of clinical and diagnostic data fed into such systems with powerful pattern-matching capabilities. AI can also eliminate manual tasks such as taking appointment notes, prescribing medications, and keeping records updated.
• Logistics, shipping, and transportation — These industries that rely heavily on location services can use generative AI to transform satellite images into map views, assisting human workers and systems alike to plan routes and get goods to destinations more efficiently.
• Manufacturing — Unplanned downtime can be financially detrimental to manufacturers of all sizes. With generative AI, manufacturers can eliminate unplanned downtime with predictive maintenance. They also can produce and test multiple product designs at once to determine which ones they should pursue for the highest profits.
• Finance — Generative AI can enhance the customer experience in retail banking by automating real-time transactions like deposits and withdrawals.
• Education — In higher education as well as primary and secondary schools, generative AI can assist teachers in creating lesson plans and tests, tracking grades and attendance, and eliminating many of the repetitive tasks of running a classroom.
• Hospitality — Hotels, restaurants, and other hospitality businesses can use generative AI to personalize guest experiences. Many tedious and repetitive tasks like booking reservations, or checking room or table availability can easily be handled by generative AI chatbots.

Cohesity and generative AI

Cohesity is committed to using artificial intelligence (AI)—particularly generative AI—to stay ahead of security threats by harnessing the power of an organizations’ data.

Cohesity is collaborating with Microsoft Azure OpenAI to give businesses more power when managing, securing, and protecting their data. With growing opportunities for AI to mitigate future threats based on risk profiles and user behavior, Cohesity is at the forefront of understanding generative, cognitive, and retrieval augmented generation AI to stop bad actors.

Generative AI is important to Cohesity’s core business of securing and managing data. Cohesity helps customers back up their entire data estate and improves cyber resilience with data isolation, threat detection, and data classification.

Cohesity provides the world’s largest organizations with the deep insights and analytics they need to improve their security postures. The Cohesity Data Cloud is a modern data security and management platform and is unique in that it is “AI-ready.” It is architected in such a way that is massively scalable, and easily searchable, and enforces granular access controls and security features to ensure the highest levels of data availability and integrity.

For example, with Cohesity’s global search capabilities, people can easily and quickly search globally across multiple workloads and data copies from a centrally managed interface. The Cohesity Data Cloud will allow AI and large language models (LLMs) to quickly answer critical business questions, while ensuring that only the right people see responses regarding the data they have access to.

Data protected in Cohesity Data Cloud is indexed and contains the specific metadata that will make utilizing that data in LLMs possible. In the same way that data is stored and able to be searched and analyzed for threats, this metadata is also AI-ready so that when a person asks questions about the data through the LLM or other power-language AI models (such as Azure OpenAI), the LLM will provide immediate and human-readable responses. By using authoritative data sources backed up on Cohesity, organizations will get more accurate and actionable responses to both user and machine queries.

Enterprise businesses, government agencies, and other organizations can safely and securely introduce AI into their cybersecurity strategies using the power of Cohesity’s Data Cloud platform to deliver comprehensive, clean, and contextual data for AI, security, compliance, and analytics initiatives.