"Artificial intelligence is triggering three revolutions at the same time, says futurologist Amy Webb at the SXSW tech trade fair and warns of the consequences. Author Chris Dixon can already see the damage to an industry today.
Amy Webb from the American Future Today Institute predicts the next “Technology Supercycle” at the South by Southwest SXSW tech trade fair. As with the invention of the railroad, the Industrial Revolution and the breakthrough of the Internet, artificial intelligence will change the "history of humanity," said Webb, who also teaches as a professor at the NYU Stern School of Business, in Austin, Texas. Their trend report has been one of the highly regarded highlights of South by Southwest for years.
She describes artificial intelligence as a flywheel that also triggers further comprehensive changes. While the railway, industrial revolution and internet only changed one large area of society and the economy, there are now three revolutions at the same time: artificial intelligence itself, biotechnology and the Internet of Things ("Connected Ecosystems of Things"). "The wave of technological innovation we expect will be so powerful and sustained that it will transform every part of society," Webb said. The rapid progress in biotechnology and the benefits of the Internet of Things are only made possible by artificial intelligence.
Warning about hallucinations and deepfakes
In her trend report, Webb also warns of the downsides, setbacks and undesirable developments of artificial intelligence. She criticizes the fact that companies like Meta or the French company Mistral AI have made their Large Language Models (LLM) freely accessible. Companies hope that this open source model will enable faster development of their language models. For example, in its early days, Twitter was also open to external developers, allowing many programmers outside the company to create applications from which Twitter itself could benefit.
But powerful language models in the wrong hands can also cause great damage. Deepfake porn is still a relatively harmless case. “What if deepfake videos or reports start a war,” Webb said. In addition, the problem of hallucination in language models has not been solved despite all the further developments of the last year. You should still not trust any text or image created with artificial intelligence without checking it.
After the Large Language Models, she sees the Large Action Models (LAM) as the next development. These models would then no longer be trained with texts from the Internet, but with data from motion sensors, cameras and minicomputers that will soon be installed in every refrigerator, every watch and every other electronic device. "While LLMs calculate what we will think next and create a text from it, the LAMs calculate what we will do next," emphasizes Webb.
Webb cites Apple's Vision Pro computer glasses as an example of a device that already collects countless data. Perhaps in the future, such a device could initiate actions that humans have not yet thought of, such as opening a website or starting a machine. But what happens if the artificial intelligence hallucinates here too?
Even more power for Apple, Google and Microsoft
At another event in Austin, tech investor and author Chris Dixon warned primarily about the problems that artificial intelligence is already causing. Because the costs of developing the LLM are so high, the already overpowering corporations such as Microsoft or Meta would further consolidate their dominance. “The fact that four or five AI companies will soon have power over the Internet is, to me, a dystopian idea,” said Dixon.
He warns that artificial intelligence will mean the downfall of many Internet companies. The business model of most websites relies on traffic, but if Google or new AI applications display the search results on their own site and no longer redirect users, these websites will collapse. Large providers like the “New York Times” could still do business with AI companies and charge money for the content with which the LLMs are trained; small websites would have no chance. “We have to think of new business models for content providers,” he warns.” [1]
1. "Die KI verändert die Geschichte der Menschheit". Frankfurter Allgemeine Zeitung (online)Frankfurter Allgemeine Zeitung GmbH. Mar 10, 2024. Von Cai Tore Philippsen, Austin
Komentarų nėra:
Rašyti komentarą