Artificial Intelligence

Basics of AI

By definition Artificial Intelligence or AI according to IBM is technology that enables computers and machines to simulate human learning, comprehension, problem solving, decision making, creativity and autonomy. One way Artificial Intelligence is used is to create original content as a response to a prompt. These AI are known as Generative AI. Generative AI relies on deep learning models that are algorithms that simulate processes of the human brain. The way these models work is to look for patterns in large datasets then using that to understand normal everyday language. Forbes classifies Generative AI into 4 groups: Large Language Models, Diffusion Models, Generative Adversarial Networks, and Neural Radiance Fields.

Large Language Models are trained on large amounts of text data which allows them to learn the relationships between words to predict what should come next. Examples of Large Language Models include ChatGPT and Google Gemini.

Diffusion Models are used in video and image generation in a process called iterative denoising. It starts with a scribble of random noise and through each step some noise will be replaced with requested traits until the desired outcome is produced. Examples of Diffusion Models are Stable Diffusion and Dall-E.

Generative Adversarial Networks are used in generating both text and images. The process involves two algorithms one being a ‘generator’ and the other being ‘discriminator’ competing against each other with the generator attempting to create realistic content and the discriminator attempting to figure out if the content is real or not. This type of AI is used in a large amount of Computer Vision and Natural Language Tasks.

Neural Radiance Fields are used to create digital 3D objects using deep learning. This method was pioneered by Nvidia by predicting elements of objects and mapping them into a 3D space. This technology is used in simulations, video games, robotics, and urban planning.

Guidelines

Data Protection

Ensure thatany sensitive or confidential information disclosed during chat sessions remains safeguarded. Refrain from sharing personally identifiable information (PII) or any delicate university-related data with AI systems.

Sensitive Data Storage Best Practices

Video Conference Sessions

Artificial Intelligence (AI) bots such as Otter.AI and Read.AI are increasingly prevalent in virtual meetings conducted via Zoom, often added by the host.

While these tools may enhance convenience, they can also pose considerable privacy and security challenges. Much like recorded meetings, data is sent to the cloud where sensitive information could potentially be accessed.

If you are in a meeting where a bot is present, do not hesitate to express your privacy concerns and request that recording be stopped. The host possesses the ability to disable these AI bots. It's important to remember that with the advent of new technology and its associated benefits, new challenges and risks can often emerge.

Ethical Risks

While Artificial Intelligence (AI) undoubtedly brings tremendous capabilities, it also introduces significant ethical risks that warrant careful consideration. Here are some key areas of concern:

Bias: AI systems learn from data available on the Internet, which may inadvertently reflect discriminatory or biased viewpoints.
Misinformation: Lacking inherent knowledge or the ability to fact-check, AI systems may propagate misinformation.
Threat Actors: The powerful capabilities of AI can be exploited by malicious actors to disseminate false information, orchestrate social engineering attacks, or create convincingly deceptive content.
Privacy Concerns: Interactions with AI often involve the exchange of personal or sensitive information. It is crucial to weigh the privacy and security implications of providing such information to an AI system, as unauthorized access or data breaches could result in the exposure of sensitive data.
Accountability Challenges: AI systems function as tools, and the onus for their output ultimately falls on the users. This creates complexities in ascertaining responsibility if the system generates harmful or unethical content. As a safety measure, always scrutinize any content produced by AI.

Vendor Due Diligence

When engaging a third-party provider for AI services, it's important to understand the license agreement. ITS should review the vendor's security policies, data safeguarding measures, and adherence to pertinent regulations. This ensures that their security practices are in harmony with CSUEB's established best practices.

Employee Training and Awareness

All employees must use best practices for interacting with AI systems like ChatGPT and upholding the University's information security standards. Any inquiries or concerns should be promptly directed to the Information Security Team at iso@csueastbay.edu.

More Information

See below for descriptions of Generative AI software that may be used on campus:

Copilot for Microsoft 365 is an AI-powered productivity tool that uses large language models and integrates your data with the Microsoft Graph and Microsoft 365 apps and services. It works alongside popular Microsoft 365 apps such as Word, Excel, PowerPoint, Outlook, Teams, and more. Microsoft 365 Copilot provides real-time intelligent assistance, enabling users to enhance their creativity, productivity, and skills.

Flip Back

Grammarly AI is an AI writing partner that understands the larger context of your email and document and with a few clicks can transform any text to the right tone, length, and clarity you need.

Flip Back

Grammarly AI is an AI writing partner that understands the larger context of your email and document and with a few clicks can transform any text to the right tone, length, and clarity you need.

Flip Back

MidJourney is an AI-driven platform that transforms text prompts into high-quality, visually stunning artwork. Whether you're an artist, designer, or someone exploring creative expression, MidJourney allows you to generate unique digital images effortlessly.

Flip Back

Apple Intelligence enhances productivity, security, and health monitoring, offering personalized, AI-driven experiences that integrate seamlessly across Apple’s devices.

Flip Back

Apple Intelligence enhances productivity, security, and health monitoring, offering personalized, AI-driven experiences that integrate seamlessly across Apple’s devices.

Flip Back

Google Gemini is an advanced AI platform by Google, combining powerful generative AI with multimodal capabilities, enabling it to handle text, images, video, and code. It enhances productivity and creativity across industries by integrating seamlessly with Google’s ecosystem, including Search, Assistant, and Workspace tools like Docs, Sheets, and Gmail.

Flip Back

Artificial Intelligence

Basics of AI

Guidelines

Data Protection

Video Conference Sessions

Ethical Risks

Vendor Due Diligence

Employee Training and Awareness

More Information

See below for descriptions of Generative AI software that may be used on campus:

Additional Resources

Campus

Legal

Tools