 | | | Good morning. In reading Ed Zitron’s latest newsletter on the “con” that is generative AI, I’ve got a lot of thoughts. I think I can sum them up in a simple idea, that we need to collectively think hard/harder about why certain actors/companies/governments are saying and doing what they are when it comes to generative AI, what their motivations are and why they should be trusted, both in their statements and their actions. | And we have to collectively think hard about how these groups are achieving, or plan to achieve, the world they are suggesting — and what might be lost, or who might get hurt, in the process. | As Five for Fighting once sang: “What kind of world do you want? Think anything.” | We all have a stake in this. | — Ian Krietzberg, Editor-in-Chief, The Deep View | In today’s newsletter: | ⚕️ AI for Good: A protein language model 🔏 New York Times greenlights use of internal AI tools 🌎 South Korea stakes a claim in the AI Race
|
| | |
| AI for Good: A protein language model |  | Source: MIT |
| While Google DeepMind’s AlphaFold was able to provide researchers with a means of predicting protein structure through the phenomenon known as protein folding, it represents only one piece of a complex, biological puzzle. | Another piece involves protein localization — the biological transportation of proteins to specialized compartments within a cell. | What happened: Researchers at MIT recently developed a machine learning model called ProtGPS that can predict protein localization. | While other research has been undertaken into the prediction of protein localization, researchers didn’t know if a protein’s localization to one of those specific cellular compartments could be predicted based on its amino acid sequence. Trained on batches of proteins with known localizations, the team’s model was able to “correctly predict where proteins end up with high accuracy.” The model is also able to predict how protein localization changes due to mutations associated with disease and illness.
| The model additionally includes a generative component, wherein it can generate novel proteins that are designed to localize to a specified location, something that could improve researchers’ ability to design novel therapeutics and drugs. | Why it matters: “My hope is that this is a first step towards a powerful platform that enables people studying proteins to do their research, and that it helps us understand how humans develop into the complex organisms that they are, how mutations disrupt those natural processes and how to generate therapeutic hypotheses and design drugs to treat dysfunction in a cell,” MIT biology professor Richard Young said. |
| | |
| | New York Times greenlights use of internal AI tools |  | Source: Unsplash |
| The news: The New York Times has approved the use of generative AI tools for its product and editorial teams, according to a Semafor report, saying that such tools could be used to write everything from code to SEO headlines and social media copy. | The Times declined a request for comment. | In addition to debuting the Times’ own generative AI tool, a model called Echo, the company also approved the use of a number of third-party systems, including NotebookLM, Google’s Vertex AI, GitHub’s Copilot coding assistant and some Amazon AI products. The Times reportedly encouraged staff to use these tools to edit their work, generate headlines, generate questions to ask in interviews and to conduct research.
| Many of the editorial use cases being offered by the company have to do with simple summarization. | The Times did note, however, that journalists shouldn’t use GenAI to significantly revise or draft an article. The technical standards for acceptable use here — i.e., how much revision is ‘significant’ revision — remain unclear. | The landscape: This comes in the midst of the Times’ increasingly tense legal battle with OpenAI and Microsoft, in which the Times has accused OpenAI of massive copyright violations associated with the training of ChatGPT. | It’s unclear what the Times’ internal chatbot is trained on, and whether that data has been licensed or is otherwise legally fair to use. | “Generative AI can assist our journalists in uncovering the truth and helping more people understand the world,” the Times’ 2024 editorial guidelines say. “The Times will become more accessible to more people through features like digitally voiced articles, translations into other languages and uses of generative AI we have yet to discover.” |
| | |
| | | Innovaccer, a healthcare data company, on Monday announced the launch of a suite of new AI agents designed to combat burnout among healthcare workers. The Trump Administration has begun firing FAA workers, an effort that is being fronted by Elon Musk. It shortly follows a major plane crash in Washington DC.
| | Why Trump’s metal tariffs won’t lead to the all-American iPhone (Wired). EU says it ‘cannot fill’ USAID funding gap as Africa seeks solutions (Semafor). Why Elon Musk offered to buy the OpenAI nonprofit (Vox). Privacy-problematic DeepSeek pulled from app stores in South Korea (Ars Technica). How studying babies' minds is prompting us to rethink consciousness (New Scientist).
|
| | |
| South Korea stakes a claim in the AI Race |  | Source: Unsplash |
| The news: South Korea on Monday announced plans to acquire 10,000 state-of-the-art GPUs in a push to remain competitive in the ever-evolving AI Race, according to a Reuters report. | The details: The country aims to secure the chips through public-private partnerships, though details concerning which chips they’ll acquire, and from which companies, won’t be settled until September. | The push represents an effort to accelerate the country’s launch of its national AI computing center. The Korean government has been discussing the “urgent need to secure high-performance AI computing resources” since last summer, and laid out the details of its national computing center plans in September of the same year.
| "As competition for dominance in the AI industry intensifies, the competitive landscape is shifting from battles between companies to a full-scale rivalry between national innovation ecosystems," South Korea's acting President Choi Sang-mok said in a statement to Reuters. | The landscape: South Korea is only the latest in an ever-lengthening line of countries to take a similar approach to the development of artificial intelligence. | The U.S. has been at the forefront of much of this infrastructure push, an effort that was further expanded by the recent announcement of Project Stargate, a venture that will apparently amount to a $500 billion investment in infrastructure by a handful of major AI companies, spearheaded by OpenAI and Softbank. | It’s a lead that the rest of the world is following; France last week announced more than $110 billion in private AI-related investments, alongside the European Union’s announcement of a plan to pour more than $200 billion into AI-related infrastructure buildouts. | Though the U.K. has yet to announce investments on par with its peers, the country has made clear that it is taking a very positive stance on AI, with an intense focus on creating an inviting environment for AI startups while pushing AI adoption in everything from the country’s government systems, to its healthcare personnel and individual consumers. | Prime Minister Keir Starmer wishes to “mainline AI into the veins of this enterprising nation.” | This is all in the midst of a tense geopolitical environment in which, beyond the tariffs that President Donald Trump is promising to ratchet into place around the globe, the U.S. and some of its rivals have put into place export restrictions on the chips, or the materials needed to build the chips, necessary for artificial intelligence. | So, why does it matter: AI is often likened to the Industrial Revolution. And that might well prove to be an apt comparison; at this stage, it’s a little too early to tell. But I have been noticing specific, fundamental differences between previous technological revolutions and this one, and South Korea’s entry into the race, a race that is now being run by countries, not just companies, is highly relevant to it. | It’s not always clear, according to Britannica, how and why agrarian societies evolved into industrial societies. The first Industrial Revolution, which began in Britain before spreading to the U.S. and the rest of the world, had a couple of key characteristics, namely, the use of new materials and energy sources (iron and coal, for example), which led to the creation of machines. Things kind of snowballed from there. After these revolutions were in swing, after factory jobs had been established, and railroads were crisscrossing the land, governments stepped in more directly to shape and feed industrialization through trust-busting, labor laws and infrastructure buildouts.
| A handful of countries, according to Britannica, “attempted to manipulate” those elements that led to industrialization in other places. These state-led industrial efforts, most prominently employed by Japan and Russia, became common among later-stage industrializers as a means of catching up. | | To the degree that we can call artificial intelligence a technological revolution, it is one that is being pointedly pushed by nations around the world, nations that are keen on enabling and opening up the necessary infrastructure, all while concurrently spearheading the adoption and use of the tech. | The scale of the revolution on hand here is one that is being, at best, partially inflated by governments at a time when efficacy and legitimate usability remain fuzzy, at a time when technological limitations and cybersecurity implications ought to prevent the adoption of the tech in high-stakes arenas. | I think of what Dr. Seena Rejal told me recently, that “the whole thing is running way ahead of policymakers. They can't really keep up with it because I don't think they really understand it at its core.” | I find it interesting that this is not a revolution that seems to be happening on its own; it’s one that nations are taking a personal hand in helping along. | A lot of the same issues — workers' rights, labor laws, monopolies, etc. — are on display here, again, as they were hundreds of years ago. But these two revolutions are distinct, at least in comparing the state of each starting line. | | | Which image is real? | | | | | 🤔 Your thought process: | Selected Image 1 (Left): | “The window patterns in the generated image was a giveaway. In the real image, you see a specific pattern in one building with variations of open/close or different curtains... In the generated image, all windows are black rectangles.”
| Selected Image 2 (Right): | |
| | |
| 💭 A poll before you go | Thanks for reading today’s edition of The Deep View! | We’ll see you in the next one. | Here’s your view on the Take It Down Act: | Nearly 70% of you support the bill and hope it gets passed. | 16% don’t love it and 1% think the states are doing a fine job. | Much-needed: | “Not sure why it didn't pass with flying colors the first time, especially when this has been affecting children as well. However, it's a relief to hear something positive is happening in legislation — we need every bit of hopeful news we can get these days.”
| Something Else: | “A good idea with good intentions, for the majority of situations it will help. But what about public figures, politicians, satire, parody, political cartoons, etc? What about tweaks to the face that it could be argued it's not the person who says it is them? Expect a LOT of court challenges.”
| What do you think about the NYT's move? | | If you want to get in front of an audience of 200,000+ developers, business leaders and tech enthusiasts, get in touch with us here. |
|
|
|
