Google recently launched a comprehensive training course on AI and machine learning, and we at Hive Creatives are delving a little deeper into the topic of AI and machine learning. Today, there are a variety of AIs powered by language models, a sophisticated type of machine learning that excels at producing natural-sounding language. This technology, often referred to as \u201cgenerative AI,\u201d raises many questions about its nature and capabilities.<\/p>\n\n\n\n
To answer frequently asked questions about generative AI, large language models, and machine learning, we reached out to Douglas Eck, a senior research manager at Google. Doug’s unique background in both AI research and the creative arts positions him perfectly to explain how generative AI works and its potential impact on technology and creativity.<\/p>\n\n\n\n
Before we dive into generative AI, it\u2019s important to understand AI in a broader context. AI, often an elusive concept, generally refers to advanced computer systems. Doug prefers to focus on \u201cmachine learning,\u201d which forms the backbone of most AI developments today. Machine learning allows computers to learn from examples, a process called training.<\/p>\n\n\n\n
Machines that are trained to learn from examples are called \u201cneural networks.\u201d For example, to teach a network to recognize an elephant, we feed it many images of elephants labeled accordingly. Over time, the model learns to identify elephants by distinguishing them from other elements in the images. Language models are a specialized type of neural network designed to understand and generate text.<\/p>\n\n\n\n
Language models predict the next word in a sequence based on extensive training data. By processing large amounts of text, these models learn context and likely word patterns. For example, if you type “Mary kicked a\u2026,” a well-trained language model would likely end the sentence with “ball” rather than a more general term like “round object.”<\/p>\n\n\n\n
Recent advances have significantly improved the performance of language models, from increasing their size to optimizing the amount of data required for training. These models already improve our daily lives through features like Smart Compose and Smart Reply in Gmail, and they also power Bard.<\/p>\n\n\n\n
Generative AI uses its learning to create entirely new content. Large Language Models (LLM), a type of generative AI, produce new text combinations that sound natural. In addition to text, generative models can create images, audio, and even video, as seen with tools like Imagen, AudioLM, and Phenaki.<\/p>\n\n\n\n
Generative AI has enormous potential for creative fields. It can automate tedious tasks, allowing creatives to focus on more innovative aspects of their work. Doug compares the impact of generative AI to the introduction of the drum machine in music, which did not replace drummers but instead created new musical genres and styles.<\/p>\n\n\n\n
While generative AI offers exciting opportunities, it also presents challenges. Doug, who has a background in education and literature, expresses concern about how teachers can assess student work in an era where AI can generate essays. This is similar to previous educational changes, such as the introduction of graphing calculators.<\/p>\n\n\n\n
Companies, including Google, have a responsibility to navigate these challenges in a thoughtful way. Google\u2019s AI development is guided by principles established in 2018, with a focus on creating helpful technology that avoids harm. This includes collaborating with the Responsible AI group and other teams working to address bias and toxicity in AI models.<\/p>\n\n\n\n
Generative AI already solves simple problems and helps with tasks like document generation. However, its future potential is even greater. Doug envisions generative AI will revolutionize creative workflows, enabling people to take on entirely new challenges with new perspectives. Through ongoing experimentation and collaboration, the full benefits of generative AI will continue to unfold.<\/p>\n\n\n\n
Generative AI, exemplified by tools like Bard, promises to reshape our creative processes and technological capabilities, offering a glimpse of an exciting future of enhanced human-machine collaboration.<\/p>\n\n\n\n
For those of you interested in AI and machine learning, you have Google courses here:<\/p>\n\n\n\n
<\/p>\n\n\n\n
https:\/\/www.cloudskillsboost.google\/course_templates\/536<\/a><\/p>\n\n\n\n https:\/\/www.cloudskillsboost.google\/course_templates\/539<\/a><\/p>\n\n\n\n https:\/\/www.cloudskillsboost.google\/course_templates\/554<\/a><\/p>\n\n\n\n https:\/\/www.cloudskillsboost.google<\/a><\/p>\n\n\n\n https:\/\/www.cloudskillsboost.google\/course_templates\/541<\/a><\/p>\n\n\n\n https:\/\/www.cloudskillsboost.google\/course_templates\/543<\/a><\/p>\n\n\n\n https:\/\/www.cloudskillsboost.google\/course_templates\/537<\/a><\/p>\n\n\n\n\n
\n
\n
\n
\n
\n