The agentic web. The machines decides, the human assumes
To understand the intervention of the set of technologies that produce artificial content, the so-called Artificial Intelligences, it is important to understand the scope of data mining and the use of deep learning, whose contribution depends on the amount of data available to each technology corporation. The more data, the greater the commercial advantage, that is, the greater the intervention over human beings, over our capacity for decision-making and will.
Regarding this use and deep learning, Karen Hao, in her essay ‘Empire of AI’, writes:
“Tech giants already had proof of the commercial potential of neural networks before the DNN-research auction. In 2009, Hinton graduate students demonstrated that this software was acceptable for speech recognition. IBM, Microsoft, and Google jumped on the bandwagon, but Google was the first technology company to achieve its commercialisation. In 2012, Google began producing neural networks and greatly improved Android's recognition capabilities.”
As Google's AI operations continued to expand, neural networks created decisive improvements for the company's gold mine: searches. The software could better match user queries with the most relevant web pages, providing users with better search results and, more importantly, offering them more personalised ads. The more Google profited and the more billions it invested in deep learning, the more the rest of the industry followed suit. Companies established themselves as the leading investors in AI research, and before long, they were setting the research agenda based on breakthroughs that could generate short-term returns.
And the key to understanding the current context:
"The deep learning overload in Silicon Valley, in its attempt to expand and consolidate large-scale monopolies, also encoded a culture among AI developers to consider everything as data that should be captured and consumed by their technologies in an attempt to make them reflect the world as much as possible. (...). The detailed digital traces of people's thoughts and ideas on social networks were simply text. People and vehicles in photos were simply objects. Surveillance was detection”.
As far as the general public is concerned, generative AI has been among us since the end of 2022 with the launch of ChatGPT by OpenAI. But that’s not true, and once again, information is being hidden from us so that we continue to be consumed and reduced to mere data. From 2012 to 2022, we witnessed what we consider the first era of AI commercialisation, laying the groundwork for the current uncontrolled revolution of generative AI. This led us to the agentic web, a model that allows generative artificial intelligence to become an extension of human beings, mimicking their reasoning processes and, therefore, modifying human behaviour, making us more dependent, more fragile, less free, and with a static global perspective heavily influenced by ideological biases.
The agentic web consists of AI systems that can perform tasks independently, make decisions, and act on behalf of users with limited human intervention, leveraging multiple models and tools to achieve their objectives.
Neópolis #4.1
Audiovisual and sound installation by LCA.
Project SOS-LCAmálaga | Promálaga Casa de Socorro (Plaza de la Trinidad, 12)
Mon 09 March, Tue 10 March, Wed 11 March, 7.30-9.00 p.m.
Free entry until full capacity is reached
Diálogo Neópolis #4.1 | Colloquium with César Rendueles
‘Redes vacías.
Tecnología catastrófica y el fin de la democracia’ (Anagrama, 2026) (Empty networks. Catastrophic Technology and the End of Democracy)
Monday 9 March, 6.30 p.m.
Promálaga Casa de Socorro (Plaza de la Trinidad, 12)
Free entry until full capacity is reached