Revealing the Secrets of Artificial Intelligence

Revealing the Secrets of Artificial Intelligence

Getting your hands on practical, everyday artificial intelligence (AI) is gradually entering the public sphere, even though it's been utilized in the business and industry world for quite some time now. The media seems to be labeling everything from graphics to content creation as the solution to various issues, but this use of AI has been giving it a bad reputation in many circles.

This misunderstanding may be due to the lack of understanding of AI, which is why my TV Tech colleague, John Footen, and I created these regular columns on this rapidly advancing technology.

This month, I'll be diving into some basics of what it takes to utilize AI (specifically "generative AI") by providing visual and textual overviews to help you understand its fundamental aspects and get started with it. Let's delve back into the past few years and start connecting the dots to get a broader perspective.

A Quantum LeapFor the first time, we're now able to ask data questions in human language and receive a unique, original answer, delivered as human text or speech. However, this new setup isn't without a catch-it requires a substantial increase in power, changes in infrastructure, and power consumption. Let me break it down for you:

• Energy: A single Google search uses roughly 100 watts of power, but the new AI solutions like ChatGPT consume far more, leading to a 600-800 times increase in power draw.
• Applications: Game theory, AI for healthcare, video understanding, human-AI interaction, ethical AI, graph/data mining, and 3D point clouds are some of the areas where AI is being applied today.

The misconception around AI may stem from its complexities and the vast amount of data required to make it work. The major players in the industry, such as Google, Oracle, Amazon, and Microsoft, are constructing massive, warehouse-like data centers to store all the required data.

Data, the Lifeblood of AIYou've probably heard the term "big data" thrown around for years, but you may be wondering what it truly means. In the early days, few realized the importance of "big data" to AI. Even now, the term isn't commonly associated with AI. But without "big data", we wouldn't have reliable, consistent, and trustworthy AI.

AI needs a large amount of data to train its systems, which is how large language models (LLMs) can generate answers based on massive data collections. However, this requires a significant investment in infrastructure, power, and real estate.

Power at the CoreIn 2021, the average IT-centric equipment rack in a data center consumed around 7kW of power. But for AI, that's not nearly enough. The report expects average rack densities to reach 20kW by 2030.

AI LandscapeThe applications for Generative AI (GenAI) are continually evolving across various industries, particularly in areas like imaging, video, text, research, 3D modeling/manufacturing, medical, speech translation, and documentation.

AI CapabilitiesGenAI can adjust the tone of content based on the intended audience. Altering the phrasing, content, and presentation can make the text, visual details, or audio more appealing to specific groups. This is often done by suggesting counterarguments, improving the emphasis on certain sections, or generating new text for audiences with unique interactive design preferences.

Web design can now be dynamically adjusted based on the user experience or leveraging data from past searches. This is commonly used on search engines to guide users towards sponsored areas of a webpage that align with both the advertiser's objectives and the user's interests.

Visual PerceptionVisual hierarchy is a technique used to capture user attention by resizing and adjusting graphic elements. This allows content to be swayed to better suit a specific user, making the media more engaging and appealing.

AI in MediaAmazon, the new owner of the 007 franchise's IP, plans to use GenAI technologies to create new stories, content, and discover new talent who can appeal more to Bond enthusiasts. Amazon's CEO, Andy Jassy, has denied any plans for AI to write a Bond movie, but we'll have to wait and see.

Depth of KnowledgeAs users interact more with the GenAI bot, they can ask follow-up questions to delve deeper into the responses. However, users should be cautious not to overcompensate with too many inquiries, or the AI might start to produce artificial answers that stray from the topic.

Researchers are concerned about the time required for AI to become reliable enough for consistent and confident use. Stay tuned for more insights on this topic in the September issue of TV Tech.

1. The misconception around artificial intelligence (AI) may arise from its complexities and the vast amount of data required to make it work.
2. In the business and industry world, AI has been utilized for quite some time now, and getting your hands on practical, everyday AI is gradually entering the public sphere.
3. This month, I'll be diving into the basics of utilizing AI (specifically "generative AI") by providing visual and textual overviews to help understand its fundamental aspects and get started.
4. A substantial increase in power, changes in infrastructure, and power consumption are required for the new AI solutionslike ChatGPT.
5. The major players in the industry, such as Google, Oracle, Amazon, and Microsoft, are constructing massive data centers to store all the required data for AI.
6. The applications for Generative AI (GenAI) are continually evolving across various industries, particularly in areas like imaging, video, text, research, 3D modeling/manufacturing, medical, speech translation, and documentation.
7. GenAI can adjust the tone of content based on the intended audience, making the text, visual details, or audio more appealing to specific groups.
8. Amazon, the new owner of the 007 franchise's IP, plans to use GenAI technologies to create new stories, content, and discover new talent who can appeal more to Bond enthusiasts.

Read also: