John Battelle's Search Blog What’s SearchGPT Really About? Moving Past the Training Data Dilemma.
OpenAI’s launch of SearchGPT appears less as a genuine competitive threat to established search engines and more as a strategic maneuver to mitigate the intense backlash regarding its data training practices. With the company facing significant pressure from publishers and governments over unauthorized web scraping, this move serves primarily to repair its public image and reassure content providers that it values their contributions. The announcement emphasizes features like prominent attribution and linking, aiming to demonstrate a symbiotic relationship with journalism. By framing the tool as a temporary prototype separate from its core generative models, OpenAI attempts to distinguish between search functionality and model training. This distinction is crucial for encouraging publishers to allow their sites to remain accessible for search indexing, thereby maintaining their supply of high-quality training data despite ongoing controversies. This development is highly relevant to the open data community as it highlights the tension between free web indexing and controlled data access. It underscores how major tech entities are increasingly adopting attribution-based models to legitimize data extraction. Understanding these shifts is vital for advocates of open data, as it reveals the growing pressure to formalize permissions and provenance, potentially impacting the free and open nature of the web by prioritizing licensed or consensual data sources.
Source: battellemedia.comPublished on 2024-07-27
Related news
- Uso de datos sintéticos a futuro no servirán a Inteligencia artificial
- To Share or Not to Share: Debates and New Approaches on Data Sharing for AI Training
- How open is open source AI – Let’s Llama
- Open Source AI Has Founders—and the FTC—Buzzing
- Google defends itself in proposed class action, says it never collected users’ personal information