Tech Beat by Namecheap – 6 October 2023

This week, in our article The Ethical Side of AI, we investigate the ethical challenges posed by the rapid adoption and integration of artificial intelligence (AI) into various sectors of society. AI has transformed from a science-fiction idea to a real-world force, driving numerous applications — from predictive algorithms to autonomous vehicles. However, its widespread use brings forth several ethical dilemmas. We highlight concerns related to data privacy, especially with AI’s massive data processing capabilities, and discuss the potential biases in AI algorithms, which can arise from biased training data. We also explore the impact of AI on the job market, with research estimating that AI-driven automation could eliminate 85 million jobs by 2025. Finally, we suggest the need for collaborative efforts between tech companies, researchers, and policymakers to ensure AI’s ethical and responsible use. 

In tech news:

• Google manipulates search results to boost revenue. In a startling revelation in the ongoing Google antitrust trial, Wired reports that the company has been altering search queries billions of times daily in numerous variations to boost its earnings. For example, a user might input “children’s clothing” only to have Google covertly change it to “NIKOLAI-brand kidswear,” delivering less accurate or desired search results. Such modifications not only enhance Google’s revenue by pushing users towards more commercial results and encouraging clicks on sponsored ads but also degrade the overall search experience for users and increase costs for advertisers.

• Hollywood writers’ strike reaches tentative agreement with production studios. The Hollywood film and TV writers’ strike has ended after 148 days. According to Vice, the Writers’ Guild of America described the deal as “exceptional—with meaningful gains and protections for writers in every sector.” A significant element of writer grievances centered around AI. The guild initially proposed forbidding the use of AI to write or rewrite material produced by writers. The Alliance of Motion Picture and Television Producers rejected this proposal but offered annual meetings to discuss advancements in AI instead. So, the use of AI in writing is not prohibited, but there are regulations on how it can be credited and trained. For example, AI-generated material cannot be given writing credit, and companies can’t require writers to use AI when producing material.

• Evernote allegedly secretly forced users onto more expensive plans. TechRadar reports that legacy Evernote customers found themselves upgraded to high-cost plans without notification or accepting new terms. These customers took to Reddit to air their grievances and find out if anyone else had experienced something similar. One user shared that their account had been changed from Evernote Plus (a plan that is no longer an option for new users) to Evernote Personal. Another user confirmed the same thing had happened to them, with a subscription increase from $79 per year for the Plus plan to $159 for Personal. The original Reddit post has since been updated to say that their account was reverted to Plus and that the change was an error. However, because of the widespread error, some speculate whether the company plans to transfer legacy users to current tiers. 

• 26% of the top 100 websites are blocking GPTBot. A recent analysis has shown that an increasing number of popular websites, including Pinterest, Indeed, and The Guardian, are blocking GPTBot, the web crawler introduced by OpenAI. According to Search Engine Land, at least 26 of the top 100 most popular websites and 242 of the top 1,000 have implemented such blocks, which marks a 250% increase from the previous month. The main reason behind these blocks is that GPTBot does not provide citations or links to its sources. However, it seems that more websites are now concerned about OpenAI scraping their data without compensation. 

• Meta’s new AI tools are both entertaining and unexpected. You’ll soon be able to use the Meta AI assistant on Instagram, Facebook, and WhatsApp. The AI’s features will probably seem somewhat familiar if you’ve played with any popular AI chatbot released thus far, with some interesting twists. According to The Verge, key features include getting answers to all your pressing questions (with Microsoft’s Bing helping to provide real-time web results), generating images through the prompt “/imagine,” and AI characters based on celebrities like Snoop Dogg and Paris Hilton. Meta has not revealed precisely what the model was trained on, apart from telling TheVerge it’s a custom large language model “based on a lot of the core principles behind Llama 2.”
  
  At the same time, Meta is coming under fire for user-generated stickers using the AI tool. As Vice reports, the tool has allowed people to create unsavory cartoon images of world leaders, child soldiers, and other racy and violent depictions, and so far, there don’t seem to be many restrictions in place. 

• Amazon’s Buy Box is central to FTC antitrust case. Amazon’s Buy Box — a section on the right-hand side of each product page featuring “Add to cart” and “Buy Now” buttons — probably doesn’t immediately strike most online shoppers as problematic. However, a report from Wired shines a light on how the box is possibly being used to manipulate both sellers and consumers, keeping prices high and degrading services. The FTC’s Antitrust complaint alleges that it’s used as an anti-competitive weapon that punishes sellers. When a product is being sold by multiple sellers, Amazon’s algorithm will select a seller to elevate in search results, with their name appearing in the Buy Box. According to the complaint, Amazon will banish users from the Buy Box and sometimes the site entirely for selling products elsewhere for lower prices. The government ultimately aims to get an injunction that will force Amazon to change how it operates. 

Previously in Tech Beat: Combatting AI bias

It’s important to recognize the significant racial and gender biases within generative AI, both image and text generative tools. In our article Addressing racial and gender bias in generative AI, we emphasize the importance of using diverse training data, testing for bias, ensuring transparency, and promoting workplace diversity to combat these issues. The future of generative AI hinges on addressing these biases to prevent the perpetuation of harmful stereotypes.

Tip of the week: Use robots.txt to block AI

It’s worth considering if you should protect your original content from being used to train AIs without your consent. 

If you decide to take this step, you can easily block OpenAI’s web crawler, GPTBot, from accessing your site. This can be done by adding just two lines of code to your website’s Robots.txt file. The Robots.txt file is a document that determines if web crawlers can crawl your website, and it is always publicly accessible. By disallowing GPTBot from crawling your site, you are effectively preventing OpenAI from using your content for training purposes.

To block GPTBot, you need to open your website’s Robots.txt file and add the following code:

User-agent: GPTBot

Disallow: /

If you only want to block certain parts of your website, you can specify the directories that should be disallowed. For example:

User-agent: GPTBot

Disallow: /directory-2/

It is important to note that it is currently unclear if web browsing versions of ChatGPT or ChatGPT plugins will be prevented by the Robots.txt file. Therefore, we recommend that you use the above methods as a precaution.

If you want to stay up-to-date with the latest tech news, be sure to subscribe to Namecheap’s blog today!