Hello and welcome to AI. In this edition … The information media are struggling with AI; Trump orders us for IA security efforts to refocus on the fight against the “ideological parts”; The distributed training wins more and more traction; The increasingly powerful AI could tip the scales towards totalitarianism.
AI is potentially disruptive for commercial models of many organizations. In some sectors, however, the threat is as apparently existential as the news sector. It happens to be the business in which I am, so I hope you will forgive a somewhat indulgent newsletter. But the news should be important to all of us, because a free functional press plays an essential role in democracy – by informing the public and helping to play the power to account. And, there are similarities between the way in which news executives are – and in a critical way, are not – adding the challenges and opportunities that the AI present that business leaders in other sectors can also learn.
Last week, I spent a day at a conference by Aspen Institute entitled “AI & News: Charting The Course”, which was organized at the Reuters headquarters in London. The conference was followed by senior executives from a number of British and European press organizations. It was held under the rules of Chatham House, so I cannot tell you who said exactly what, but I can relay what has been said.
Tools for journalists and publishers
Information managers talked about using AI mainly in internal oriented products to make their teams more effective. AI helps to write the optimized major titles of the engines and to translate the content – allowing potential organizations reach new audiences In places, they have not traditionally served, but most of them pointed out by keeping humans in the loop to monitor precision.
An editor described using AI to automatically produce short articles from press releases, releasing journalists for more original reports, while maintaining human publishers for quality control. Journalists also use AI to summarize documents and analyze large data sets – such as government documents and satellite imagery – allowing investigative journalism that would be difficult without these tools. These are good use cases, but they translate into a modest impact, which makes work flows more effective.
Below or from top to bottom?
There has been an active debate among press chiefs and technicians of the editorial room on the question of whether press organizations should adopt an upward approach – by going beyond the generative tools of AI in the hands of each journalist and publisher, allowing these people to execute their own data analysis or “Ambrance code” The widgets fed by AI to help them in their work, or if the efforts should be top to bottom, management prioritizes projects.
The ascending approach has merits – it democratizes access to AI, authorizes front -line employees who often know the points of pain and can often identify good cases of use before high -level leaders can, and releases limited AI developer talents to spend only for more important, more complex and potentially more strategically important projects.
The disadvantage of the ascending approach is that it can be chaotic, which makes it difficult for the organization to guarantee compliance with ethical and legal policies. It can create a technical debt, with tools built on the fly which cannot be easily maintained or updated. A publisher feared to create a two -level editorial room, some publishers embracing the new technology, and others who are late. Associate also does not guarantee that the solutions generate the best return on investment – a key consideration because AI models can quickly cost expensive. Many have called for a balanced approach, although there was no consensus on how to reach it. According to the conversations I have had with leaders in other sectors, this dilemma is familiar in all industries.
Beware of Jéopardizer Trust
Press outfits are also cautious about the construction of public -oriented AI tools. Many have started using AI to produce summaries of articles that can help occupied and increasingly impatient readers. Some have built AI chatbots that can answer questions about a particular and narrow subset of their coverage – like stories On the Olympic Games or climate change– But they tended to qualify them as “experiences” to help report readers that the responses are not always accurate. Few went further in terms of content generated by AI. They fear that the hallucinations produced by AI-AI including confidence in the accuracy of their journalism. Their brands and companies ultimately depend on this confidence.
Those who hesitate will be lost?
This prudence, although understandable, is itself a colossal risk. If the press organizations themselves do not use AI to summarize the news and make it more interactive, technological companies are. People are turning more and more to AI search engines and chatbots, including perplexity, Openai Chatppt and Google Gemini and “IA glimpses” that Google now provides in response to many research, and many others. Several news leaders of the conference said that “disintermediation” – the loss of a direct link with their audience – was their greatest fear.
They have reasons to worry. Many press organizations (including Fortune) Depend at least in part the Google research to call on the public. A recent study by toll– The one who sells software that helps protect websites from web robots – I came across click prices for Google AI’s overviews was 91% lower than that of a traditional Google search. (Google has not yet used an overview of AI for news requests, although many think that this is only a matter of time.) Other studies on clicks from chatbot conversations are also appalling. Cloudflare, which also offers to protect publishers from news from web scratch, found that Optai had scratched an information site 250 times for each reference page, he sent this site.
So far, press organizations have responded to this potentially existential threat thanks to a mixture of legal decline – the New York Times continued Openai for copyright violations, while Dow Jones and the New York Post continued the perplexity– and partnerships. These partnerships involved seven -digit license agreements for news content. (Fortune To a partnership with perplexity and prorate.) Many conference leaders said that license agreements were a means of generating content income that technological companies had probably already “stolen” anyway. They also considered partnerships as a way to build relationships with technological companies and to exploit their expertise to help them create AI products or train their staff. No one has seen relationships as particularly stable. They were all aware of the risk of becoming too dependent on income from AI licenses, having been burned before when the media industry has let Facebook become a major traffic engine and advertising revenues. Later, this money practically disappeared overnight when the meta-PDG Mark Zuckerberg decided, after the 2016 American presidential election, to support news in people’s flows.
A Ferrari powered by AI was pulled in a horse cart
Managers admitted have needed to build direct audience relationships that cannot be disintermediary by AI companies, but few had clear strategies to do so. An expert in the conference declared frankly that “the information industry does not take AI seriously”, focusing on “increasing adaptation rather than structural transformation”. He compared the current approaches to a three -step process which had “a Ferrari powered by AI” at both ends, but “a horse and a cart in the middle”.
He and another advisor to the media industry urged press organizations to move away from the structuring of their approach to news around “articles”. Instead, they have encouraged the directors of news to think about the ways of which the source material (public data, interview transcriptions, documents obtained from sources, raw video sequences, audio recordings and archive news) could be transformed into a variety of outings – peaks, short -term video tastes, bubble -point summaries, or yes, Public on the fly by AI generative technology. They also urged press organizations to stop thinking about the production of news as a linear process, and starting to think more like a circular loop, perhaps that in which there was no human in the middle.
One person of the conference said that press organizations should become less island and more closely examine ideas and lessons from other industries and how they adapted to AI. Others have said that this might require startups – perhaps incubated by the press organizations themselves – to launch new commercial models for the age of AI.
The stakes could not be higher. Although AI poses existential challenges to traditional journalism, it also offers unprecedented opportunities to develop the scope and potentially reconnect with the public who “deactivated the news” – if the leaders are daring enough to reinvent the news in the AI era.
With that, here is more news from AI.
Jeremy Kahn
jeremy.kahn@fortune.com
@Jeremyakahn
Correction: The edition Tuesday of last week of Eye we have Misplying the country where Trustpilot has its registered office. It’s Denmark. In addition, a news in this edition has poorly identified the name of the Chinese startup behind the Model of the Viral model of the AI. The name of the startup is a butterfly effect.
This story was initially presented on Fortune.com
