Operation Retweet: Meet Venezuela’s AI journalists
September 4, 2024
Good morning. The AI Boom hit a bump in the road Tuesday. Shares of Nvidia fell 9% and shares of Wall Street’s broader chip index retreated some 7%.
Nvidia has now fallen around 13% since its good, but not good enough earnings report last week.
— Ian Krietzberg, Editor-in-Chief, The Deep View
In today’s newsletter:
AI for Good: Europe’s planetary defense system
Source: ESA
Next month, the European Space Agency’s (ESA) self-driving Hera spacecraft will begin its two-year journey to survey a nearby asteroid system.
The details: Hera has been in development for several years now. The spacecraft will operate similarly to an autonomous vehicle, analyzing input from a variety of sensors to build and navigate a model of its surroundings.
Where other deep space missions have a remote driver on Earth, Hera’s autonomy will (hopefully) allow it to get within 200 meters of the asteroid’s surface for detailed scientific observation.
Hera’s post-impact survey of the asteroid represents the “world’s first test of asteroid deflection.”
Why it matters: The ESA said that, once proven, the autonomous technology being tested by Hera can and will be applied to a variety of other missions, acting as the building block for “low-cost planetary probes into deep space.”
Change The Way You Think About Hydration
Our body’s water reserves decline with age, making it even harder to stay hydrated. On top of that, even minor dehydration can affect how you feel. One study found losing just 2.8% of your body weight in fluid can impair memory and thinking! So it’s essential you remember to replenish fluid levels every day.
Enter: NativePath Native Hydrate. This 3-in-1 hydrating powerhouse isn’t your average electrolyte drink. With electrolytes, amino acids, and 11 additional nutrients, this combination can help combat dehydration, inflammation, poor circulation, UTIs, and muscle weakness. It’s also sugar-free with no artificial ingredients.
Try this drink mix from NativePath risk-free for up to 44% OFF (PLUS: a free gift for The Deep View readers!).
Clearview AI fined $33.7 million for ‘illegal’ practices
Source: Clearview
Clearview AI — a facial recognition platform that works with law enforcement and intelligence organizations — was fined $33.7 million by the Dutch Data Protection Authority (DPA) for illegal data collection.
Clearview says its law enforcement database has more than 50 billion images of people; the Dutch DPA said that the company scrapes these photos from the internet without consent.
Dutch DPA chairman Aleid Wolfsen acknowledged the role of facial recognition in criminal investigation and detection, but said that it should never be performed by “a commercial business.” According to Wolfsen, such technology should only be used by law enforcement in “highly exceptional cases,” and under the supervision of data protection authorities.
Going further: Wolfsen said that since Clearview breaks the law, it is therefore illegal to use its services.
The Dutch DPA said that Clearview violates Europe’s General Data Protection Regulation (GDPR) in several ways, adding that it “should never have built the database” in the first place.
Wolfsen said the Dutch DPA is now investigating whether it can hold Clearview’s directors personally responsible for the violations. It added that since Clearview did not object to the decision, it cannot appeal the fine.
Context: The company’s been embroiled in controversy for years, recently settling a class action lawsuit that similarly took issue with its massive facial database.
“Facial recognition is a highly intrusive technology, that you cannot simply unleash on anyone in the world,” Wolfsen said. “This is not a doom scenario from a scary film. Nor is it something that could only be done in China.”
Clearview’s response: In a statement, Clearview’s chief legal officer said that it doesn’t operate anywhere in the EU and so isn’t subject to the GDPR. He called the decision “unlawful, devoid of due process and unenforceable.”
The Dutch DPA acknowledged that it doesn’t operate in Europe, but said that since its database includes images of Dutch citizens and other European citizens, it is subject to European law.
Lock in a 6.9%* yield with a Bond Account on Public
On Public, there’s a simpler way to take advantage of the highest corporate bond yields since 2009. Meet the new Bond Account:
Invest your cash in a Bond Account
Lock in your 6.9%* yield
Earn interest payments 20x a year
Take advantage of historic yields
If you act now, you can lock in a 6.9%* yield before the Fed potentially cuts rates in September and beyond.
Turn your cash into bonds
With as little as $1,000, you can invest in a portfolio of ten investment-grade and high-yield bonds.
Get regular fixed-income payments
While you are invested in a Bond Account on Public, you'll receive regular interest payments.
Public is the all-in-one investing platform where you can build a multi-asset portfolio of stocks, options, bonds, crypto, and more.
AI agent startup Agency raised $2.6 million in pre-seed funding.
Government contract startup Pryzm raised $2 million in pre-seed funding.
Want to become an AI consultant? 200+ students have already started and scored their first client in as little as 3 days. Request early access to The AI Consultancy Project.*
The sneaky way Big Tech is acquiring AI unicorns without buying the companies (CNBC).
Facebook and Instagram ran ads violating Kenyan election law, new report reveals (Rest of World).
Intel's Dow status under threat as struggling chipmaker's shares plunge (Reuters).
The US Navy is going all in on Starlink (Wired).
Big Tech ‘Clients’ of Jacob Wohl’s Secret AI Lobbying Firm Say They've Never Heard of It (404 Media).
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.
Study: The impacts of a lack of transparency in LLMs
Source: Unsplash
Large language models (LLMs) are trained on increasingly massive datasets, diverse collections that get combined and recombined in model-specific shuffles.
The origin of much of the data employed to train LLMs gets lost in the process, and according to a new study from MIT, the impacts of this are vast.
The details: Ethically, unknown data sources are likely to contain, at the very least, biases, though the problem also poses legal concerns regarding copyright and licensing.
But, according to the researchers, it could also damage a model’s performance; miscategorized datasets could lead someone to train a model for a task with data that wasn’t designed for that task.
A team of MIT researchers launched an audit of more than 1,800 text datasets in an effort to improve transparency. They found that around 50% had information containing errors and 70% didn’t include some licensing information.
So, what: The researchers then developed a Data Provenance Explorer tool that auto-generates summaries of datasets, focusing on creators, licenses and sources.
“These types of tools can help regulators and practitioners make informed decisions about AI deployment, and further the responsible development of AI,” MIT Professor Alex Pentland, a co-author of the paper, said in a statement.
Operation Retweet: Meet Venezuela’s AI journalists
Source: Connectas
Since July, the people of Venezuela have been exposed to two new news anchors online, known as ‘La Chama’ and ‘El Pana.’ The anchors have been reporting on the geopolitical goings-on of Venezuela following the presidential elections of July 28.
They aren’t real.
The details: The two anchors are Generative AI avatars, launched as a part of a new project — called Operation Retweet — by Colombia-based news organization Connectas.
The operation serves as a strategy to avoid and evade both censorship and repression.
Connectas says that, although the avatars aren’t real, the data and information they are providing is, adding that it has been verified by a dozen local and international media outlets.
The landscape: “The use of Artificial Intelligence is not a fad,” according to Connectas. It is being “used as a tool to protect journalists from persecution and increasing repression after the elections.”
Following the presidential elections, foreign journalists have been deported and at least nine Venezuelan journalists have been arrested, according to Reporters without Borders. Eight of them remain imprisoned on “trumped-up charges and in worrying conditions.”
The journalists are facing between 12 and 20 years in prison for charges including “terrorism.”
The project’s goal is to provide people with necessary, relevant information while protecting the journalists who would otherwise be visibly providing that information.
Connectas director Carlos Huertas said that, with AI avatars, there’s no one to arrest.
“We decided to use artificial intelligence to be the 'face' of the information we're publishing," Huertas told Reuters, "because our colleagues who are still out doing their jobs are facing much more risk."
He called Connectas’ use of genAI a “mix between technology and journalism," saying that it’s needed to "circumvent persecution and increasing repression."
This is a great example both of the necessary intentionality required to employ generative AI, as well as the exceptional, rare instances where it can — and perhaps even should — be employed.
The realistic anonymity generative AI offers is, in many cases, a threat that is primed for the spread of disinformation. In other, less intentional cases, it is a threat primed for the unknowing distribution of false, biased or otherwise hallucinatory information.
To employ it in a way that guarantees the veracity, fairness and legitimacy of the information being espoused requires plenty of caution and tons of human oversight — with that caution and oversight, the avatar certainly acts as, in this case, a necessary mask.
The high stakes of its deployment aren’t lost on me, either. It was in the face of repression and arrests that Connectas tried to find another way, not to jump on a trend, not to trim its staff, but to do its job.
This is very much not the same as the GenAI social media influencers you might see (as I do) pushed, every day, through Twitter’s wonderful (torturous) algorithm.
Which image is real? |
🤔 Your thought process:
Selected Image 2 (Left):
“Look closely at the top window in the fake picture and there’s something about the depth of it in the wall of the house and shadows that is obviously off.”
Selected Image 1 (Right):
“Shingles seem a little off on #2 though the inclusion of the waterfowl makes it a tough choice.”
💭 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 how AI should impact labor:
Around 40% of you said that it should only be allowed to be used in specific ways and in specific professions. Close to a third of you would like it to enable a four-day workweek.
20% of you said it should replace all labor, so long as we get a Universal Basic Income.
What do you think about Clearview & commercial facial recognition databases? |
*Public disclosure: All investing involves the risk of loss, including loss of principal. Brokerage services for US-listed, registered securities, options and bonds in a self-directed account are offered by Public Investing, Inc., member FINRA & SIPC. Cryptocurrency trading services are offered by Bakkt Crypto Solutions, LLC (NMLS ID 1828849), which is licensed to engage in virtual currency business activity by the NYSDFS. Cryptocurrency is highly speculative, involves a high degree of risk, and has the potential for loss of the entire amount of an investment. Cryptocurrency holdings are not protected by the FDIC or SIPC.
Alpha is an experiment brought to you by Public Holdings, Inc. (“Public”). Alpha is an AI research tool powered by GPT-4, a generative large language model. Alpha is experimental technology and may give inaccurate or inappropriate responses. Output from Alpha should not be construed as investment research or recommendations, and should not serve as the basis for any investment decision. All Alpha output is provided “as is.” Public makes no representations or warranties with respect to the accuracy, completeness, quality, timeliness, or any other characteristic of such output. Your use of Alpha output is at your sole risk. Please independently evaluate and verify the accuracy of any such output for your own use case.