Issue link: https://digital.copcomm.com/i/1512899
www.postmagazine.com 33 POST NOV/DEC 2023 OUTLOOK O O OUTLOOK O OUTLOOK ARTIFICIAL INTELLIGENCE oday we're seeing an acceleration of AI advancements, particu- larly generative AI, and its potential to revolutionize media and entertainment workflows. We see AI impacting every area, from planning and preproduction, to post production, distribution and even audience engagement. AI-driven visualizations let artists and directors plan shots quickly and eŸ- ciently. AI can help editors streamline post production by automating tasks like video trimming, scene detection and color grading. Generative AI-powered search and indexing tools can enhance content discoverability and audience engagement. Our software and solution partners have been thoroughly integrat- ing AI-backed features into their appli- cations. These include industry leaders like Adobe, Autodesk, Avid, Blackmagic Design, Foundry and Maxon. These features do what AI does best: simplify redundant tasks, provide more insights about your content and make previous- ly-complex procedures more accessible.— In some cases, these AI features are processed in the cloud. Others can be processed locally using the integrated AI functionality of modern GPUs, like Nvidia's RTX line — featured in Dell Precision workstations, which are recom- mended for M&E pros. Software provid- ers recognize the value of local process- ing for minimizing latency with the larger data sizes of video and audio files, and oˆsetting the considerable costs of oper- ating these features in the cloud. Cloud operating costs for software vendors are so substantial that they are moving to put some of these features "on the meter" for end users. As custom- ers come to use and depend on these innovative and impressive capabilities, it will not be surprising if they ask for more local processing to minimize their costs. Some may request this simply because they do not want any part of their video or audio data leaving the safety of their company firewall. Along with the rapid expansion of generative AI in our industry, there are concerns about privacy, intellectual prop- erty protection, safety and governance, particularly with public cloud-based generative AI models. Industry leaders in the market, including Dell, are quickly committing to responsible AI approaches that aim to protect customers' privacy and content. But as the regulatory and governance standards for responsible AI emerge, our leading customers have recognized an approach to leverage the benefits of generative AI while reducing their concerns. Public and open-source generative AI large language models (LLMs), such as Stable Diˆusion and Meta's Llama 2, can be operated and fine-tuned pri- vately for specific tasks, and operated entirely behind the firewall. This does not necessarily mean huge racks of GPU- powered servers that are often associat- ed with generative AI. It can also include workstations that have the GPU power to handle these LLMs. They leverage techniques like LoRA and RAG, which are eˆective at reducing the computational weight of LLMs without dramatically compromising their accuracy. While LLMs may be more suitable in the near term for larger media enterpris- es because it can scale eˆectively on smaller machines, Dell can meet custom- ers wherever they are in their AI journey. Customers can choose to use their favorite apps with innovative new AI features, or scale their business transfor- mation with full-scale AI deployments with the confidence that they are in control of their content. AI is bringing rapid and exciting innovations to the market, and while caution is merited, everyone should be exploring the opportunities to use AI to transform the way they do business. A.I. IN MEDIA & ENTERTAINMENT WORKFLOWS BY MATT ALLARD DIRECTOR OF ALLIANCES & SOLUTIONS DELL TECHNOLOGIES DELL.COM T for posting to social media. That's time saved on manual reframing while expe- diting delivery. While AI tools may produce quick results, they also must be accurate to truly elevate the visual quality of a project and actually save time. With the release of DaVinci Resolve Studio 18.5 this year, accuracy took center stage with additional DaVinci Neural Engine-based features across diˆerent post disciplines. For example, editors can transcribe audio within clips to search for media based on narrative content, or quickly generate subtitles for timelines with the automatic speech-to-text feature. DaVinci Neural Engine AI can analyze and automatically sort audio clips based on classification, and colorists can use Relight FX to add virtual lighting to a—scene. At Blackmagic Design, it's always been about empowering creativity, so that's where we hope to see the future of AI go, with people using the AI tools in DaVinci Resolve Studio to increase eŸciencies in their workflows so they can then focus more of their time and energy on creativity. Adobe Premiere's speech-to-text transcription feature.

