| | | Good morning. As much as AI might seem like magic, it isn’t, for a litany of reasons that we’ve explored over the past weeks. But there is another aspect to the reality of AI that we have yet to dive into — the entire data-labeling ecosystem behind AI. | So, here we go. Read on for the full story. | In today’s newsletter: | 📟 Nvidia announces next-generation AI chip platform 📄 Paper: The importance of terminology in AI 🧱 Is deep learning hitting a wall? 🕰️ The workers that make AI tick accuse U.S. big tech giants of ‘modern-day slavery’
|
| | |
| | Nvidia announces next-generation AI chip platform |  | Jensen Huang, CEO of Nvidia (Nvidia). |
| Nvidia CEO Jensen Huang announced Sunday that the company’s next-generation AI chip platform will be called “Rubin” and will become available in 2026. | He didn’t provide much in the way of details regarding the coming chip family — we’re still waiting for Nvidia to launch the next-generation iteration of its Blackwell family of chips — but he did say that, from now on, Nvidia plans to release a new AI chip family every year.
| You can watch clips from his keynote address here. | | Paper: The importance of terminology in AI |  | Photo by Sven Brandsma (Unsplash). |
| Much of the AI industry comes down to terms, phrases and shaky definitions. The term “AI” itself, according to some researchers, is a hype term, based more in mythology than reality, implying the creation of an alien intelligence rather than statistical algorithms. | A recent paper from Dr. Gina Helfrich (of the University of Edinburgh) argues that the term “frontier AI” should be retired. | Key Points: | She argues that the term allows its proponents to focus the public conversation on existential (hypothetical) risks, rather than the active social, psychological and environmental harms that are caused and perpetuated by large language models today. “Profit. Danger. Outer space. Progress. These are the connotations of ‘frontier AI.’ It should be obvious that ‘frontier AI’ is an exercise in AI hype, given these connotations,” she said.
| Helfrich dives deep into the linguistic history of AI and a series of terminologies that have sprung up around it, as well as their sources. You can read the full paper here. | Why semantics in AI matter. As computer scientist Jaron Lanier wrote last year: “The easiest way to mismanage a technology is to misunderstand it.” Inaccurate terminology is the first step on the road to misunderstanding. | | Together with Artisan AI | Hire Your First AI Employee: Ava, An AI BDR | | Ava automates 80% of your outbound cycle so your team can focus on high-leverage tasks like taking calls and closing deals. She operates alongside your team in the Artisan platform, which consolidates: | 300M+ High-Quality B2B Contacts Automated Lead Research & Enrichment Optimized Email Deliverability AI Playbooks that Replicate Human Research & Email-Writing Workflows Multi-Language Support in 40+ Languages
| Get in touch and supercharge the productivity of your BDR team. | | Is deep learning hitting a wall? |  | Photo by Markus Spisky (Unsplash). |
| More than two years ago, cognitive scientist Gary Marcus wrote an essay titled ‘Deep learning is hitting a wall’ that incited the derision of the deep learning community. In the article, he argued that deep learning (the machine learning subset that relies on multi-layered neural networks to simulate the workings of the human brain) has reached a point of diminishing returns. | “With its toxic language, bias, unreliability and tendency towards fabricating misinformation,” Marcus argued that the field requires a paradigm shift.
| Elon Musk on Saturday reposted a meme that had circulated back in 2022 in response to Marcus’ original article; Marcus responded by saying that his two-year-old points still stand. | | Gary Marcus @GaryMarcus Incredible. @elonmusk is here mocking my claims of diminishing returns and @sama just let slip that GPT-4 goes back at least to November 2022.We are talking 18 months and > $50 billion without a major bump, and no real solution to hallucinations, unreliability, poor reasoning… x.com/i/web/status/1… C Chaitanya @nutanc So OpenAi had already trained GPT-4 by the time they released ChatGPT?So basically we have not had a better model(even by OpenAI) for the last 2 years? x.com/bilawalsidhu/s… Jun 1, 2024 452 Likes 73 Retweets 37 Replies |
| “For all the daily claims of ‘exponential progress,’ and for all the tens of billions that have been invested since, reliability is still a dream, and of course toxicity and hallucinations are still with us,” Marcus wrote. “Worse, we are no closer to a solution to the alignment problem now than we were then.”
| Zoom out: A recent paper — explained simply in this clip by Computerphile — pushed back against the idea that the key to general intelligence is more data. The study found an “exponential need for training data which implies that the key to ‘zero-shot’ generalization capabilities under large-scale training paradigms remains to be found.” | “The suggestion is that you can keep making your models bigger, but we are soon about to hit a plateau where you don’t get any better,” Mike Pound, a computer science researcher at the University of Nottingham, said.
|
| | |
| | 💰AI Jobs Board: | Director of Data Systems: Central Lending · United States · Lakeland, FL · Full-time · (Apply here) Photonics Architect: Ayo · United States · Boston, MA · Full-time · (Apply here) Data Scientist: Stripe · United States · New York · Full-time · (Apply here)
| | 📊 Funding & New Arrivals | | | 🌎 The Broad View: | American tech giants are building data centers in Chile. Local activists are pushing back (Rest of World). Amazon is expanding its drone delivery services (CNBC). Increasing use of renewables also improves air quality (The Guardian). Google agrees to delete billions of data records collected from users in incognito mode to settle a lawsuit (Wired).
| *Indicates a sponsored link |
| | |
| Together with Gamma | | Whether they’re for work or school, the worst part about presentations is making them. | Even when existing options work right, the results are usually far from exciting. | Gamma aims to change that. | Gamma’s presentation app has AI solutions built directly into the architecture, meaning that you — without any design or coding skills — can seamlessly and quickly create unique, customizable, professional-looking presentations … all with nothing more than a simple prompt. | You can join Gamma’s 18 million other users — for free — in dramatically upping your presentation game right now, right here. |
| | |
| | The workers that make AI tick accuse U.S. big tech giants of ‘modern-day slavery’ |  | Created with AI by The Deep View. |
| I’m sure you’ve heard it said that the most important thing in the creation of AI models centers around the training data. | It’s true – but that’s only half the story. In order for these models to recognize elements of the world users will ask about, the data (in image and text pairs) that powers them needs to be accurately labeled. This process is called “Supervised Learning.” | Other learning methods do exist, including unsupervised learning, though many labs will incorporate supervised fine-tuning into their model training process. Supervised learning tends to result in more accurate models. ChatGPT, for instance, was trained using supervised fine-tuning, in which “human AI trainers provided conversations in which they played both sides — the user and an AI assistant.”
| But the practice of these real humans sifting through digital metric tons of data, according to a recent letter from a group of these workers, is abusive. | The letter, signed by 97 data labelers, AI workers and content moderators who work for Meta, OpenAI and Scale AI in Kenya, called on President Joe Biden to hold U.S. tech companies accountable for their “unlawful” behavior overseas. “U.S. Big Tech companies are systemically abusing and exploiting African workers,” the letter reads. “In Kenya, these U.S. companies are undermining the local labor laws, the country’s justice system and violating international labor standards. Our working conditions amount to modern-day slavery.”
| The work, they say, is “mentally and emotionally” draining. It involves “watching murder and beheadings, child abuse and rape, pornography and bestiality, often for more than 8 hours a day,” and often at less than $2 per hour. | They added that many of them live and work with post-traumatic stress disorder as a result of the work & that union-busting is rampant in the sector. | “Kenya has high levels of unemployment and the tech boom has created much-needed new jobs here, particularly for young people who are the majority of digital workers,” the letter reads. “We need these jobs, but not jobs at any cost.”
| OpenAI didn’t respond to a request for comment. Meta declined comment. |
| | |
| | | |  | Image 1 |
| Which image is real? | |  | Image 2 |
|
| | |
| | Expo: Monetize AI expertise. Gotalk.ai: Transform text into voiceovers with AI. Vizard.ai: An AI-powered video editing tool.
| Have cool resources or tools to share? Submit a tool or reach us by replying to this email (or DM us on Twitter). | *Indicates a sponsored link |
| | |
| SPONSOR THIS NEWSLETTER | The Deep View is one of the world’s fastest-growing newsletters, adding thousands of AI enthusiasts a week to our incredible family of over 200,000! Our readers work at top companies like Apple, Meta, OpenAI, Google, Microsoft and many more. | If you want to share your company or product with fellow AI enthusiasts before we’re fully booked, reserve an ad slot here. |
| | |
| One last thing👇 |  | Christie Wilcox @NerdyChristie |  |
| |
This Third Thumb is *so cool*. I am unbelievably jealous of the folks who got to try it out! That story and more of the best in @ScienceMagazine and science in this edition of #ScienceAdviser: science.org/content/articl…
Video credit: Dani Clode Design and the Plasticity Lab |  | | | May 30, 2024 | | | | | | 2.1K Likes 486 Retweets 76 Replies |
|
| That's a wrap for now! We hope you enjoyed today’s newsletter :) | What did you think of today's email? | | We appreciate your continued support! We'll catch you in the next edition 👋 | -Ian Krietzberg, Editor-in-Chief, The Deep View |
| | |
|
|
