{"id":822927,"date":"2025-01-30T00:11:50","date_gmt":"2025-01-30T06:11:50","guid":{"rendered":"https:\/\/newsycanuse.com\/index.php\/2025\/01\/30\/how-publishers-are-choosing-which-llms-to-use\/"},"modified":"2025-01-30T00:11:50","modified_gmt":"2025-01-30T06:11:50","slug":"how-publishers-are-choosing-which-llms-to-use","status":"publish","type":"post","link":"https:\/\/newsycanuse.com\/index.php\/2025\/01\/30\/how-publishers-are-choosing-which-llms-to-use\/","title":{"rendered":"How publishers are choosing which LLMs to use"},"content":{"rendered":"
\n
\n
\n

By Sara Guaglione<\/a><\/span> \u00a0\u2022\u00a0 January 30, 2025<\/span> \u00a0\u2022<\/p>\n<\/div>\n

\n

<\/p>\n

\n Ivy Liu <\/p>\n<\/p><\/div>\n<\/p><\/div>\n

\n

Publishers that want to experiment with using generative AI technology to build products and features like creating chatbots and analyzing data have to evaluate which large language models best fit the bill.\u00a0<\/p>\n

And it turns out that one of the biggest factors in these evaluations is how easy it is to integrate an LLM into their companies\u2019 <\/strong>tech systems \u2014 such as different product suites and content management platforms <\/em><\/strong>\u2014 according to conversations with three publishing execs. That often means choosing the LLMs owned by companies with which they already have enterprise technology or content licensing agreements.<\/p>\n

\n

For example, a spokesperson at one publisher \u2013 who asked to remain anonymous \u2014\u00a0 told Digiday their company isn\u2019t experimenting with an array of different LLMs and is primarily <\/em><\/strong>using OpenAI\u2019s models. Their company has a content licensing deal with OpenAI that followed a successful project to build a chatbot using OpenAI\u2019s GPT model. The publisher has continued to use GPT for other needs, like productivity tools.<\/p>\n

Another publishing executive, who spoke on the condition of anonymity, said they are using OpenAI\u2019s GPT model for internal use instead of other LLMs, simply because they have a content deal with OpenAI.\u00a0<\/p>\n

And a third publishing exec, who also requested anonymity, said their company is using LLMs owned by companies like Microsoft and Google because it is already paying to use their enterprise software, including Microsoft Office 365 and Google Workspace.<\/p>\n

This makes it \u201cvery easy for us to integrate it into our full development ecosystem,\u201d the third publishing exec said. As a customer of the Microsoft Office 365 product suite, the company can \u201cintegrate [Copilot] natively into the tools that we already use, like Outlook, Excel, Powerpoint [and] Word,\u201d they added.<\/p>\n

Nate Landau, chief product and technology officer at TheSkimm, said using technology offered by companies they already have existing agreements with means they don\u2019t have to build their own solutions.<\/p>\n

\u201cSome of our partners, like [data storage company] Snowflake, offer their own AI integrations and we prioritize those over building solutions from scratch where applicable,\u201d he said.<\/p>\n

And because TheSkimm doesn\u2019t have an exclusivity deal with an AI tech company, they have the flexibility to work with different models, Landau added.<\/p>\n

\u201cGiven the rapidly evolving nature of this space and the varying strengths and weaknesses of different models, we haven\u2019t centralized on a single provider,\u201d he said. \u201cFor each use case, we evaluate multiple providers to ensure the best fit.\u201d<\/p>\n

Some other factors media companies take into account when evaluating which LLMs to use include the cost, the performance and quality of the models\u2019 outputs and privacy and security considerations.<\/p>\n

\u201cIt is extremely important to us that the models are not trained on our users\u2019 information or inputs and that any interactions with AI products are safe and protected,\u201d said Vadim Supitskiy, Forbes\u2019 chief digital and information officer.<\/p>\n

Testing performance<\/strong><\/h2>\n

LLMs have to work well for publishers\u2019 use cases in order for all of this to be worth their time and money.<\/p>\n

When choosing which LLM to use for different processes, TheSkimm runs side-by-side tests of the models and compares their outputs \u201cto ensure they align with our brand voice and editorial standards,\u201d Landau said. That\u2019s because the major differences between these models are their \u201cvoice, tone and accuracy across various use cases,\u201d he added.\u00a0<\/p>\n

For example, Claude has a \u201csofter, more natural tone\u201d and is \u201cparticularly accurate,\u201d which is why Landau prefers it for tasks where response voice and tone are especially important. But for tasks requiring less creativity \u2014 such as working with TheSkimm\u2019s datasets \u2014 he prefers models like Meta\u2019s Llama \u201cfor their reliability in providing accurate, actionable responses and avoiding hallucinations.\u201d\u00a0<\/p>\n

However, as LLMs continue to rapidly evolve, these distinctions are \u201cbecoming increasingly subtle,\u201d Landau said.<\/p>\n

TheSkimm uses LLMs for three things in particular: data analysis, audience acquisition and experimentation, according to Landau. These models help TheSkimm surface key trends and cohorts from its data, and help create audience segments to target messaging and products, such as for its shopping and commerce businesses and for sponsored content.<\/p>\n

The third publishing exec said their company uses Google Gemini and its services for product development, such as building chatbots. For internal efficiency tools the publisher uses Microsoft Copilot. The company is currently evaluating Google Gemini to integrate the AI features into Google Workspace<\/a> (which includes products like Gmail, Docs and Meet, among others).<\/p>\n

Cost<\/strong> considerations<\/strong><\/h2>\n

Mark Howard, chief operating officer at Time, said he considers the financial incentives to work with one LLM over another (Time has a content licensing deal with OpenAI<\/a>) as well as the opportunity to be involved in future product development to benefit the business by having a seat at the table to help shape new products that can be beneficial for publishers.<\/p>\n

\u201cSome of [these deals] are more about being part of new marketplaces that they\u2019re developing. And to me, the part that\u2019s the most interesting \u2014 companies looking to really build something that doesn\u2019t currently exist. As they\u2019re building those, I would rather be a part of that,\u201d Howard said. Time also has deals with Perplexity and ProRata, to participate in the former\u2019s ad revenue share program and the latter\u2019s per-use compensation structure.<\/p>\n

TheSkimm is currently using a combination of open-source models<\/a> (such as Meta\u2019s Llama and French-based Mistral) and commercial products, including Anthropic\u2019s Claude and OpenAI\u2019s GPT models. <\/p>\n

The company weighs the \u201cpros and cons\u201d of paying fees to access private LLMs\u2019 APIs and the potentially more cost-efficient option of hosting their own open-source models, Landau said.<\/p>\n<\/div>\n

\n

https:\/\/digiday.com\/?p=566832<\/p>\n<\/p><\/div>\n<\/p><\/div>\n

\n

\n More in Media<\/span>
\n <\/h3>\n<\/p><\/div>\n<\/div>\n

Read More<\/a>
\n Sara Guaglione<\/p>\n","protected":false},"excerpt":{"rendered":"

By Sara Guaglione \u00a0\u2022\u00a0 January 30, 2025 \u00a0\u2022 Ivy Liu Publishers that want to experiment with using generative AI technology to build products and features like creating chatbots and analyzing data have to evaluate which large language models best fit the bill.\u00a0 And it turns out that one of the biggest factors in these evaluations<\/p>\n","protected":false},"author":1,"featured_media":822928,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[25101,1461,46],"tags":[],"class_list":{"0":"post-822927","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-choosing","8":"category-publishers","9":"category-technology"},"aioseo_notices":[],"_links":{"self":[{"href":"https:\/\/newsycanuse.com\/index.php\/wp-json\/wp\/v2\/posts\/822927","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/newsycanuse.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/newsycanuse.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/newsycanuse.com\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/newsycanuse.com\/index.php\/wp-json\/wp\/v2\/comments?post=822927"}],"version-history":[{"count":0,"href":"https:\/\/newsycanuse.com\/index.php\/wp-json\/wp\/v2\/posts\/822927\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/newsycanuse.com\/index.php\/wp-json\/wp\/v2\/media\/822928"}],"wp:attachment":[{"href":"https:\/\/newsycanuse.com\/index.php\/wp-json\/wp\/v2\/media?parent=822927"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/newsycanuse.com\/index.php\/wp-json\/wp\/v2\/categories?post=822927"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/newsycanuse.com\/index.php\/wp-json\/wp\/v2\/tags?post=822927"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}