Market Metrics

Total Crypto Market Cap: down 2.8% to $2.42T
Total AI Sector Market Cap: down 4.4% to $20.2B

Top Movers (24 hours):

📈 Sensay (SNSY): up 10.7% to $0.001697
📈 Aleph.im (ALEPH): up 8.5% to $0.2971
📈 Carbon Browser (CSIX): up 4.6% to $0.05797

Weekend News

🟠 The United States Department of Homeland Security established the the Artificial Intelligence Safety and Security Board, which includes many prominent AI leaders such as Sam Altman (OpenAI). We tend to agree with Joseph’s take here.

🟠 Virtual protocol teased a partnership with the OpenTensor foundation.

🟠 Node AI released one-click node deployment.

🟠 io.net founder Ahmad Shadid issued a statement, claiming that the platform had undergone a sybil attack from bad actors gaming the points program. The statement addressed the reasons for the exacerbated count of available GPUs on the platform and detailed the actions taken to fix it.

🟠 Deep Link, an AI cloud gaming protocol, announced its public sale is starting soon.

🟠 Carbon Browser teased a Browse-to-Earn program with various rewards such as a CSIX staking yield bonus and Carbon Browser Pro.

Sound Bites 🎧

In this week's edition of Sound Bites, we listened to Delphi's podcast with Andrej Radonjic (0xdrej) from Grass, a decentralized web scraping protocol built on Solana. The conversation revolved around the inequalities in web scraping and how Grass aims to revolutionize the industry. Let's break down the points of discussion. 👇

Web Scraping and the Implications

Web scraping is the process of extracting data from websites. As more companies have become aware of the value of their data on the Internet, it is becoming increasingly common for them to deploy techniques to limit the scraping ability by restricting the IP addresses of data centers or employing techniques such as honey-potting which feeds the scraper incorrect data (More on this later).

To get around this, companies have started to install software development kits containing a web scraper onto people's devices without them knowing. This web scraper will leverage people's devices to scrape web data continuously in the background. This means that users unknowingly provide their resources (CPU and bandwidth) to these companies without compensation. Free apps downloaded to your phone or the screen saver apps saved on smart TVs are notorious for containing this web scraper software.

As companies become aware of the value of their web data, they employ techniques like rate limiting and honeypotting to protect their information. Rate limiting involves restricting the amount of data that can be scraped, while honey potting involves feeding fake data to detected scrapers. For example, according to Andrej, Target previously scraped Walmart's web store daily to check product prices and price their offerings more competitively. Walmart caught onto this and began showing incorrect prices to Target's scrapers, preventing them from using Walmart's site to their advantage.

The consequences of honey potting web scrapers also have dire implications for the development of AI due to the concept of data poisoning. Data poisoning, a problem where incorrect or manipulated data is fed to AI models during training, can lead to incorrect or biased outputs. This can have severe consequences, as AI models trained on manipulated data can influence the actions of millions of people. For instance, an AI model trained on poisoned data could be used to swing an election or manipulate public opinion on certain topics, raising serious concerns about the integrity of AI.

Just as companies have manipulated their ranking on search engines through Search Engine Optimization (SEO), there is a risk that people or organizations could manipulate the training data for AI models for their own gain. For example, a shoe company could pay an AI company to change all mentions of shoes in the training data to reflect their own offerings so that any output would only mention that particular company's shoes. This becomes particularly concerning when considering the persuasive power of AI models like ChatGPT, which can be more convincing than regular web searches.

Grass’ Role

Grass aims to address these issues by enabling users to have control over the web scraping process and benefit from their participation. With 2 million users now scraping and cleaning web data, Grass is working towards building an internet-scale web index that is accessible for AI model development. By providing data on specific areas and constantly updating AI models with real-time information, Grass can help build highly accurate, narrow AI models while ensuring that users are fairly compensated for their data.

As the demand for AI training data continues to grow, platforms like Grass could play a crucial role in democratizing access to data while mitigating the risks of data poisoning and manipulation. Grass is currently running a points campaign, which is entirely free to participate in and will likely lead to an airdrop.

To participate, you will need a referral code, accessible here!

AI Art of the Day

Enjoy these beautiful pieces!

Exclusive Report: Top 5 AI Cryptocurrencies to Watch 2024

🌵 Access the report here! 🌵