AI's Generative Models Flood the Web with Spam

In the rapidly evolving field of AI, staying informed can be challenging. To help, here's a roundup of recent developments in machine learning, including noteworthy research and experiments that have flown under the radar.

This week, SpeedyBrand, a company leveraging generative AI to generate SEO-optimized content, emerged from stealth mode with support from Y Combinator. While its funding ($2.5 million) and customer base (around 50 brands) remain modest, it highlights the impact of generative AI on the web.

Generative AI models are now making it cheaper and easier to create low-quality content, as highlighted by James Vincent of The Verge. Newsguard, a company specializing in vetting news sources, has exposed numerous ad-supported websites with generic names that disseminate misinformation generated by AI.

The rise of such sites poses a problem for advertisers. Newsguard's report identified nearly 400 instances of ads from 141 major brands appearing on 55 of these misleading news sites, which exploit programmatic advertising systems.

However, the concerns go beyond advertisers. As pointed out by Kyle Barr of Gizmodo, a single AI-generated article can generate substantial engagement, potentially diverting advertising revenue away from legitimate sources. The cost of generating such content is often significantly lower than the revenue it generates, creating an attractive proposition for bad actors.

The question then becomes: what is the solution? Barr suggests that search engines and ad platforms should tighten their grip and penalize those who abuse generative AI. However, given the rapid pace of the field and the scalable nature of generative AI, keeping up may prove challenging.

While spammy content is not new, the current wave is distinct due to the low barriers to entry in terms of cost and time investment. Vincent maintains an optimistic outlook, suggesting that the prevalence of AI-generated junk content could drive the development of better-funded platforms. However, the decisions made now regarding generative AI and its impact on the web will have far-reaching consequences.

Here are other noteworthy AI stories from recent days:

1. OpenAI introduces GPT-4: OpenAI has officially launched GPT-4, its latest text-generating model, available through its paid API. GPT-4 improves upon its predecessor, GPT-3.5, by accepting image and text inputs and performing at a "human level" on professional and academic benchmarks. However, it is not without flaws.

2. Controlling "superintelligent" AI: OpenAI has established a new team, led by Chief Scientist Ilya Sutskever, to explore ways to steer and control "superintelligent" AI systems.

3. NYC enforces anti-bias law: New York City has begun enforcing a law requiring employers to submit algorithms used for recruitment, hiring, or promotion to an independent audit and disclose the results publicly.

4. Valve's stance on AI-generated games: Valve has clarified its position on AI-generated assets in games, stating that its policy is evolving and not an outright rejection of AI-generated content.

5. Humane's Ai Pin unveiled: Humane, the startup founded by former Apple designers Imran Chaudhri and Bethany Bongiorno, has revealed its first product, the Ai Pin. This wearable gadget features a projected display and AI-powered capabilities, representing a unique form factor.

6. Concerns about EU AI regulation: Prominent tech founders, CEOs, VCs, and industry leaders in Europe have signed an open letter to the EU Commission, warning that stringent AI regulations could impede innovation and cause Europe to miss out on the generative AI revolution.

7. Deepfake scam on the rise: A deepfake video of Martin Lewis, a UK consumer finance champion, promoting an investment opportunity supposedly backed by Elon Musk has surfaced, highlighting the increasing prevalence of AI-generated deception.

8. AI-powered pleasure companion: Lovense, known for its remote-controllable sex toys, has introduced the ChatGPT Pleasure Companion. This beta product, integrated into the company's remote control app, offers users erotic stories generated by the companion based on selected topics.

Next Article