Read Before Monday #17

Welcome to this weeks edition of #RBM :) Have you heard that Dell is struggling to enforce a return-to-office policy? Who would ever thought that it would be a good idea, eh? Dell employees resistance highlights the value of remote work's personal and financial benefits and questions the necessity of physical office spaces given effective remote collaboration tools. In another sphere - the AI one -, Tim O'Reilly explores the legal and ethical challenges of using copyrighted content for AI training, advocating for new business models that ensure fair compensation for creators. He suggests a cooperative approach between AI companies and creators, though the implementation of pay-per-output models presents challenges. Meanwhile, Netflix's Anne Aaron and her team continue to revolutionise video streaming with advanced encoding techniques like per-shot encoding and the AV1 codec, significantly improving efficiency and setting industry standards. On the financial front, FICO and major credit bureaus engage in monopolistic practices - what a shock, right? - that increase consumer costs and burden individuals with correcting credit report errors, highlighting systemic issues within the credit industry. Finally, the Flexipede, created by Tony Pritchett in 1967, stands as a pioneering piece of computer animation, showcasing the technical achievements of early computer graphics and reflecting the innovative spirit of that era.

___

Nearly Half of Dell’s Workforce Refused to Return to the Office - discusses Dell's recent struggle to enforce a return-to-office (RTO) policy. Introduced in February 2024, the policy required employees to classify as either remote or hybrid. Remote workers were told they would not be eligible for promotions or new roles, while hybrid workers had to be in the office at least 39 days per quarter. Despite these measures, nearly half of Dell's U.S. full-time workforce chose to remain remote, valuing personal and financial benefits such as increased leisure time and savings on commuting costs. This resistance reflects a broader trend in the tech industry, where employees question the necessity of physical offices given the effectiveness of remote collaboration tools and the global dispersion of teams. Some employees even considered seeking job opportunities at companies without strict RTO mandates.

• My take: Return to the office is back. For everyone. I admire Dell employees that said "or else" to returning to a situation that doesn't help anyone. If you work with a computer, 90% of that time can be done anywhere in the world - yes, there are country restrictions because of tax purposes - but you don't need to be at a desk, in an office, where a lot of the time is spent commuting and not having a healthy lifestyle. For the remaining 10% and for necessary F2F work, yes, absolutely, go to the office. People need interactions and talk to each other, rather being on video all the time. I've been working from home for the past 10 years (since I moved to the UK), and it always worked for me. Going back to the office every day or most days of the week means that I would probably go "elsewhere".

___

Tim O'Reilly discusses the legal and ethical challenges of using copyrighted content for AI training. It explores the need for new business models and institutional frameworks that ensure fair compensation and recognition for content creators. O'Reilly advocates for a cooperative approach where AI companies and content creators work together, possibly through mechanisms like pay-per-output rather than pay-per-training, to create a sustainable and mutually beneficial ecosystem. This could lead to a new golden age for both AI development and copyright-based businesses.

• My take: Tim O'Reilly - that one, yes - is an amazing author and publisher, so if he writes about AI, I want to know it! I understand that his area - and most creative areas - are deeply affected by AI and GenAI; first because they're training on top of them, then they're being used to produce content that competes with them. I've been working with GenAI since it was launched and at Hitachi Digital Services , we have a Responsible AI approach of what we do internally and projects with our customers, where we consider all the ethical and legal issues before anything we do. But this needs to be a collaborative approach within the industry that builds AI / GenAI and the one that is being used to train the models. We need to figure out a way to protect users and organizations, plus have all the right data to train models properly. Feels like it's the chicken and the egg situation for fair compensation and someone needs to take the first step. I don't think that Pay-Per-Output is a good idea; it would require that people would know what they want when they ask for something.

___

Inside Netflix’s Bet on Advanced Video Encoding: discusses how Anne Aaron and her team at Netflix have revolutionised video streaming by optimising encoding techniques to improve visual quality and reduce bandwidth usage. Key innovations include per-shot encoding, which adjusts settings for individual segments within a video to handle complex scenes more effectively, such as those in "The Crown" with heavy smoke or "Barbie Dreamhouse Adventures" with high glitter content. Additionally, Netflix's adoption of the AV1 codec has significantly reduced bitrate requirements, and their ongoing work on future codecs promises even further improvements. Netflix's collaboration with institutions like the University of Southern California to develop and open-source video quality assessment algorithms has also set industry standards.

• My take: Netflix has always been at the forefront of video capabilities, processing and infrastructure. I have huge respect for their technical teams to pull this amount of tech into a product that is effective, fast and works. Now they're at it again and reinventing codecs, for the better - hopefully. I'm still not sure who consumes this "high glitter content", but the per-shot encoding allows to have different encoding settings on the same video, which is mind blowing. Then the work they're doing with AV1 to reduce 30% bitrate with the same quality reminds me of the first time I heard of MP3 that could compress WAV into just 10% of its size. Finally, they're expanding to gaming too and that’s awkward to understand, but at least isn't podcasting ;)

___

"Inside FICO and the Credit Bureau Cartel" gets into the monopolistic practices of FICO and the major credit bureaus—Equifax, Experian, and TransUnion. It highlights how FICO's pricing strategies set the stage for credit bureaus to increase their charges, exacerbating the cost burden on consumers. Consolidation over the decades has led to a triopoly that dominates the credit market, with FICO scores becoming integral to various financial transactions. Despite regulatory efforts, the responsibility for correcting credit report errors still falls on consumers, often leading to significant personal and financial repercussions.

• My take: oh boy oh boy! This is a juicy article, isn't it? Basically, is embodies everything that is wrong with credit, credit companies and regulators in the US - and in some other countries too. They're not there for consumers, but rather for themselves. When companies consolidate in the market they operate, we the consumers, lose. Fixing your credit rate is still a thing left for the consumer itself and that might affect their financial standing and access to credit, employment and even doing contracts in the UK (services, housing, etc).

___

The Flexipede, created by Tony Pritchett in 1967, is a pioneering piece of computer animation, notable for being one of the first character animations to tell a story and include a soundtrack. Developed using the BL-120 microfilm recorder at Culham Laboratory, the animation was produced by programming the Atlas Computer with Fortran. This early animation project involved intricate leg movements defined by arrays and coordinated through data cards, representing a significant technical achievement in the field of computer graphics and animation.

• My Take: You thought there was no retro post this week, right? :) I'm not THAT old to use a flexipede, but my dad used it when he was at Uni doing his degree in Civil Engineering, so I have a few boxes of punch cards around and grow up always with the idea or mixing up the cards and change the order - or have I already? :) The point is that this is a very cool piece of tech. Imagine creating programs on paper cards that you had to load into a machine. If you think that a millisecond refresh time is a lot for a page, imagine the hours waiting... Like the time you had to wait for a cassette to load, or a cartridge - I'm from that generation. This is a great story of computer animation done with the tech available back in those days.

___

This Week in GenAI

For those that missed the live this week, here it is #TWIGAI, where Marco Silva , Ben West and I talked about some of these articles.

• Meet Figma AI: Empowering Designers with Intelligent Tools.
• Amazon Is Investigating Perplexity Over Claims of Scraping Abuse
• Finding GPT-4’s mistakes with GPT-4
• Slack Combines ASTs with Large Language Models to Automatically Convert Unit Tests
• From the NY Times: What the Arrival of A.I. Phones and Computers Means for Our Data
• Here in London, the premiere of movie with AI-generated script cancelled after backlash
• Still in the UK, Amazon-Powered AI Cameras Is Used to Detect Emotions of Unwitting UK Train Passengers
• Finishing with UMG that will offer Voice-Cloning tech to their Artists