Copilot might soon get more Microsoft AI models, less ChatGPT presence
As per The Information, employees at Microsoft's AI unit recently concluded the training of 'a new family of AI model' that are currently in development under the 'MAI' codename. Internally, the team is hopeful that these in-house models perform nearly as well as the top AI models from the likes of OpenAI and Anthropic.
Under the leadership of its AI chief, Mustafa Suleyman, Microsoft is launching this initiative to trim down its dependence on OpenAI and develop its own AI stack for Copilot applications. The developments are not surprising.
In the last week of February, Microsoft introduced new small language models called Phi-4-multimodal and Phi-4-mini. They come with multi-modal capabilities, which means they can process text, speech, and vision as input formats, just like OpenAI's ChatGPT and Google's Gemini.
These two new AI models are already available to developers via Microsoft's Azure AI Foundry and third-party platforms such as HuggingFace and the NVIDIA API Catalog. In benchmarks shared by the company, the Phi-4 model is already ahead of Google's latest Gemini 2.0 series models on multiple test parameters.
'It is among a few open models to successfully implement speech summarization and achieve performance levels comparable to GPT-4o model,' Microsoft noted in its blog post. The company is hoping to release its 'MAI' models commercially via its Azure service.
Aside from testing in-house AI models for Copilot, Microsoft is also exploring third-party options such as DeepSeek, xAI, and Meta. DeepSeek recently made waves by offering a high performance benchmark at a dramatically lower development cost. It has already been adopted by numerous companies and recently claimed a theoretical cost-to-profit ratio of over 500% on a daily basis.
Aside from developing its own AI models to replace OpenAI's GPT infrastructure for Copilot, Microsoft is also reportedly working on its own reasoning AI models, too. This would pit Microsoft against OpenAI products such as GPT-o1 as well as Chinese upstarts such as DeepSeek, both of which offer reasoning capabilities.
Apparently, the work on an in-house reasoning model has been expedited due to strained relationships between Microsoft and OpenAI teams over technology sharing. According to The Information, Suleyman and OpenAI have been at odds over the latter's lack of transparency regarding the intricate workings of its AI models such as GPT-o1.
Reasoning models are deemed to be the next frontier for AI development, as they offer a more nuanced understanding of queries, logical deduction, and better problem solving capabilities. Microsoft also claims that its Phi-4 model delivers stronger language, mathematical, and visual science reasoning chops.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Forbes
an hour ago
- Forbes
Winning AI Search: How To Optimize For Discovery In A Generative World
Dani Nadel, President and COO, Feedvisor. Gone are the days of traditional SEO, when ranking on page one guaranteed success. Search is undergoing its most radical transformation since Google became a verb. That once-simple search box is now a generative system: anticipating needs, synthesizing answers and reshaping how consumers discover, evaluate and buy. Today's AI search agents, think ChatGPT, Perplexity and Google's Search Generative Experience (SGE), go beyond serving links. They interpret intent, scan sources and deliver personalized, conversational answers that often replace the need to browse or click. How Search Is Fragmenting Across Platforms Google still drives nearly 90% of global search traffic, but signs of fragmentation are emerging: • Younger Audiences Shift: Gen Z and millennials are turning to TikTok, Instagram and YouTube—visual, conversational spaces where search feels more organic and less link-driven. Google usage among Gen Z declined 25% compared to Gen X, signaling a generational pivot. • AI Accelerates The Zero-Click Era: AI overviews now appear in over 13% of Google queries, pushing users toward instant answers. In 2024, up to 60% of Google searches ended without a single click; over 75% on mobile. This is no accident. Tools like Gemini 2.5, powering 'AI Mode,' are keeping users on-platform with synthesized, multimodal responses. • AI-Native Platforms Gain Ground: By early 2025, 40% of U.S. internet users had tried tools like Perplexity or Google's SGE. ChatGPT commands an 80% share and over 500 million users. Gen Z leads, with 82% saying they use AI search tools at least occasionally. • Search Becomes Mid-Funnel: Only 13% of shoppers begin with Google when they know what they want. Most use it to compare, validate or explore options, not to discover. The Rise Of Agentic Commerce By 2026, Gartner predicts that search engine volume will drop by 25%, losing market share to AI chatbots and agents, tools that answer questions and take action. Early efforts like Google's Project Mariner preview this world: Bots that browse, compare and purchase on behalf of users. This shift fundamentally changes visibility dynamics. Instead of crawling and ranking pages, AI curates and selects what it deems the best answer. The implications are clear: • Fewer Clicks: Instant answers reduce site traffic. Visibility must happen upstream. • Higher Trust Threshold: Structured, credible and current content is prioritized. • Intent Over Keywords: AI rewards clear, purposeful language aligned with user goals, not keyword stuffing. • Performance Signals Count: Metrics like dwell time, conversions and reviews influence how AI weighs content. Ranking isn't the goal anymore; being chosen is. My company's internal testing shows that selection hinges on clarity, structure and completeness, traits that help AI systems quickly understand and trust your content. These systems favor content designed for rapid comprehension: Well-organized pages with clear headings, concise bullets and repeatable phrasing that mirrors real user queries like 'What does this do?' That trust deepens when your brand appears consistently across marketplaces and third-party sources. The more visible you are, the more credible you become. To earn visibility, you need to earn AI trust. These systems are the new gatekeepers to discovery. While you can't game them, you can design for them. How AI is Rewriting Marketplace Discovery This disruption isn't limited to search engines. AI is reshaping discovery and shopping across Amazon, Walmart and TikTok, often before shoppers reach a search bar. AI agents now surface marketplace listings directly in response to queries. Whether a consumer starts on ChatGPT, Google SGE or TikTok, they see curated product options, bypassing traditional in-platform search. And with tools like Google's AI Mode, agents can monitor availability, track prices, send alerts and guide users toward purchase decisions. Marketplaces are not just destinations; they are sources fueling AI-driven shopping journeys. Amazon vs. Walmart: Two Paths To AI Discovery Amazon, one of the many marketplaces my company supports, and Walmart both recognize AI product discovery as the future, but their approaches differ: • Amazon: Seems to be mostly operating a closed AI ecosystem, controlling discovery within its platform. From Rufus to product-ranking algorithms, Amazon's AI prioritizes first-party signals like pricing, availability and ratings, and auto-summarizes key product detail page details such as specs and usage scenarios. Success will require strong optimization across product pages, competitive pricing and campaign performance. • Walmart: Appears to be betting on the open web. Walmart is making its catalog AI-readable and bot-accessible, preparing for agents like OpenAI's Operator to shop across platforms. Its strategy centers on cross-platform visibility through structured data, review syndication and influencer content that ensures listings are indexed and recommended by AI. Both reward clarity, credibility and consistency, but draw from different signals: Amazon prioritizes internal data, while Walmart scans across the web. How To Optimize for AI-Driven Marketplaces AI ranks what it understands, what performs and what it trusts. Whether you're trying to rank for ChatGPT, Google SGE, Amazon or Walmart, the principles of AI-native content apply: • Prioritize relevance and clarity. Use benefit-led, conversational language. Prioritize titles, bullets and descriptions that highlight key features and differentiators to influence both AI and buyer decisions. • Answer AI-driven prompts. Frame content around real questions like "What problem does this solve?" or "How does it compare?" Mirror how users phrase queries. • Design for AI comprehension. Use structured formatting, semantic tags, concise sentences and complete fields like specs, features and use cases to support parsing and analysis. • Elevate visual content: Apply metadata, alt text, lifestyle imagery, charts and short, benefit-driven video above the fold to boost engagement signals. Well-structured visuals aid shopper decision-making and help AI extract information. • Strengthen engagement signals. Post-purchase prompts and sampling to drive reviews. Keep Q&A active. • Distribute influence off-platform. Support mobile, voice and off-site journeys that drive traffic to PDPs. What Winning in AI Discovery Requires AI is now the first impression. Whether through a chatbot, generative engine or voice assistant, your brand is being evaluated by algorithms before shoppers even begin to browse. Winning requires evolving from traditional SEO to an AI-native discovery strategy, one designed for people and the AI agents guiding their choices. Amazon and Walmart are charting different paths, but both demand smarter content and sharper signals. What you surface, and how clearly you communicate it, will determine what gets seen, what gets clicked and what ultimately gets bought. Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
Miami Herald
an hour ago
- Miami Herald
As companies embrace AI, these leaders offer tips to make it better
In 2025, it seems like just about everyone is talking about artificial intelligence and its myriad potential uses. "I have never seen any buzzword so quickly adopted," said Leila Rao, founder of Decatur, Georgia-based organizational change consultancy AgileXtended. Rao helps guide companies through a variety of organizational changes, including navigating AI. Increasingly, Georgia businesses are using AI to simplify and streamline various tasks, such as scheduling, copywriting, directing phone calls and more. In their first survey on the topic, Duluth-based Moneypenny, a phone answering and customer contact service, found that 64% of companies are now either using or considering using AI as of May to "drive efficiency, enhance decision-making and supercharge growth." Additionally, 25% of companies said they are "fully embracing" AI, though that definition was left up to the interpretation of survey respondents. But is AI actually helping workers out as much as advertised? Rao and a top executive from Moneypennysay that the current trend may be falling short of ideal, but improvement is within reach. What's the draw? Forms of AI have been used for years in many workplaces for data analytics, sales and other practices. But generative AI, powerful tools such as ChatGPT that can create original content, sounds, software code and assist in things like scientific research, has thrust AI into mainstream use. One of the unique - and attractive - things about AI as a tool is its variability, according to Rao. "With these new tools, there is no standardization because they are, almost by definition, interactive," Rao said. Countless users have also been enticed by the speed of AI - both the speed with which it can complete tasks and with which new versions and capabilities are developed. This has been reflected in the quickly-developing pervasiveness of AI in business, according to Moneypenny North America CEO Richard Culberson. Outsourcing tedium to machines Some of the top areas where it's being adopted are marketing, content creation and analytics, Moneypenny found. According to Culberson, many of the tasks AI is being used for are under the umbrella of customer interactions. The most cited benefits of embracing AI in business included cost and time savings, productivity efficiencies and better decision-making. "The business is interested, I think, mostly in how AI can fix some of the fundamental friction or drag in organizations," Rao said. Certain complex tasks are typically better left to human workers, or at having their involvement, both Rao and Culberson agreed. But Rao noted that complexity can be deceptive. Scheduling is typically done at a low level of anorganization, but can be complex in practice because of many moving parts, she noted. Culbersonsaid he favors methods that rely on human-AI collaboration to complete tasks. He cited copywriting, whether that be a human writing something for the AI to polish or vice versa, as an example of a good hybrid use. Is it actually helping? While manycompanies are diving into the world of AI and companies are making huge investments in expanding capabilities, the question remains whether these implementations are actually showing benefits or are just something for C-suite executives to flaunt. "Everybody's saying AI can give you more efficiency … the problem is, that's not translating for the recipient," Rao said. According to Rao, while higher-ups may be excited about their company's AI models, the actual workers on the ground aren't experiencing much growth or otherwise seeing significant benefits from the implementation of these systems. "More often than not, what I am seeing is the people who rave about AI outcomes are not connected to the work itself," Rao said. She isn't hating on AI as a whole though. On the contrary, she loves it - when it's used to its strengths. "I rely on AI more than I ever thought I would," she said. "But it's not a replacement for my people." The problem is, according to Rao, that many companies are failing to take the people into consideration. "There's not enough shared understanding, between AI and the people who do the work, to make meaning happen," Rao said. Part of the disconnect is emotional, Rao says; the increasing use of and conversation around AI is triggering fear among workers that they'll be replaced or deprioritized. In part, these fears are stoked by miscommunications from higher-ups failing to have a two-way conversation with these workers. Additionally, there tends to be a gap in understanding between what executives and decision-makers think and hear, versus what it's actually like doing the work on the ground, which can lead to misguided assumptions. "Despite their best efforts, leaders tend to talk to other leaders more often than they talk to and listen to their staff," Rao said. Moneypenny found that 50% of companies surveyed say they need better guidance on how to effectively implement AI, something both Rao and Culberson's companies are working to provide. These companies need help understanding how to blend "the empathy and the depth from a human perspective, as well as the technical competencies and efficiency and cost savings you get from AI," Culberson said. Part of the problem is where the AI is actually being implemented within organizations. "Very few people in business are looking across all of their departments and seeing how it can be leveraged in the right way," Culberson said. Addressing key concerns According to Moneypenny's findings, the three biggest concerns around AI are job loss, data security and customer dissatisfaction. To deal with the first problem, Moneypenny promotes a "human-centric" approach. Rao also encourages actually making AI use and development a process involving collaboration across the entire organization. "It has to be an iterative process that feels very much like a conversation," Rao said. She urges companies to have conversations - and lots of them - with people who haven't been represented in the rooms where decisions are made. As part of the process, she suggests focusing on training AI as a coach or assistant, rather than a tool to actually complete tasks - similar to Culberson's hybrid approach. Both agreed that AI should support - not replace - workers. "People who will win in the use of AI will start to elevate their own people and not eliminate their people with it," Culberson said. Having "good, productive conversations" and learning with clients are also vital for assuaging customer concerns, according to Culberson. "People want to know they're being taken care of," he said. This is an area that may see a lot of growth and advancement over the next 12-18 months. Copyright (C) 2025, Tribune Content Agency, LLC. Portions copyrighted by the respective providers.
Chicago Tribune
an hour ago
- Chicago Tribune
Liz Shulman: Will Big Tech transform school into an AI video game?
'Why am I learning AI if it's going to eventually take my job?' one of my students asked me at the end of the school year. 'I don't know,' I said. 'I wonder the same thing about mine.' Students are off for the summer, but Big Tech is working hard pitching its brand to schools, marketing its products to students as 'homework buddies' and 'personal tutors' and to educators as 'teaching assistants' and 'work pals' while undermining the entire field of education and sending out a sea of mixed messages. We all have reason to worry. The dizzying pace at which artificial intelligence has infiltrated schools and dominated the discourse within education has left the classroom a battleground of contradictions. Our fears aren't hyperbolic. Schools in Texas and Arizona are already using AI to 'teach' kids with educators as mere 'guides' rather than experts in their content area. Last year, one of my seniors told me she preferred AI to her teachers 'because I can talk to AI in the middle of the night, but my teachers don't email me back until the next morning.' In May, Luis von Ahn, CEO of the foreign language education app Duolingo, said: 'It's just a lot more scalable to teach with AI than with teachers.' Schools will exist mostly just for child care. And President Donald Trump's April 23 executive order calls for the use of AI in schools, claiming the 'early exposure' will spark 'curiosity and creativity.' This pressure isn't only coming from the White House. Education websites have uncritically embraced AI at a stunning pace. Edutopia used to highlight resources for teaching literature, history, art, math and science and instead is dominated now by AI 'tools' marketed to burned-out, overworked educators to save time. EdTechTeacher and call AI 'knowledgeable colleagues' and 'friendly buddies,' shifting away from teachers' specific subject areas. If this isn't dizzying enough, when we educators are directed or forced to use AI in our teaching, we're criticized when we do. What's really happening in the classroom is this: Teachers are unable to teach the problem-solving skills kids will need as they grow up and are blamed when an entire generation is outsourcing their imaginations to Big Tech. No wonder test scores have plunged, and anxiety and depression have risen. Yet in glossy AI advertisements paid with the billions of dollars Big Tech is making off schools, the classroom is portrayed as student-centered spaces where kids engage with personalized technology that differentiates better than teachers as though it's just another school supply item like the pencil cases on their desks. The kids know it. When I teach grammar, students want to use Grammarly. When we read a book together, they say ChatGPT can summarize it for them in seconds. When I teach them any part of the writing process, they list the dozens of AI apps that are designed to 'write' the essay for them. Students readily admit they use AI to cheat, but they're constantly getting messages to use their 'writing coach,' 'debate-partner' and 'study buddy.' It's always been an uphill battle for educators to get kids to like school. It's part of the profession. 'It's our job to push students, and it's our students' job to resist,' a mentor told me when I was a new teacher. 'In the middle,' he continued, 'therein lies learning.' Wherein lies learning now? Will school become a video game packaged as, well, school? If educators don't teach writing, we're told we're not teaching students how to communicate. If we don't teach reading, we're told we're not teaching them how to think critically. If we don't teach them business skills, we're told we're not preparing them to enter the workforce. Now we're being told if we don't teach them AI, we're not preparing them for their future that consists of what, exactly? The future that's poised to steal their jobs? At the end of the school year in my freshman English class, we read Erich Remarque's novel 'All Quiet on the Western Front.' I asked my ninth graders to choose passages that stood out to them. Many of them chose this one: 'We are forlorn like children, and experienced like old men. We are crude and sorrowful and superficial — I believe we are lost.' They noticed the alienation the soldiers feel from themselves. I wondered if it's how they felt, too — estranged from their own selves. Ironically, their discovery showed the whole point of reading literature — to understand oneself and the world better and to increase one's capacity for empathy and compassion. As my mentor teacher told me decades ago, therein lies learning. Our kids have become soldiers caught on the front lines in the battle for education, stuck in the crossfire of Big Tech and school. The classroom — a sacred space that should prioritize human learning, discovery and academic risk-taking — has become a flashpoint in America, and our kids are in the center of it. I recently finished reading 'The Road Back,' Remarque's sequel to 'All Quiet on the Western Front.' The novel dramatizes the ongoing alienation of the soldiers once they've returned home from war. 'Why can't you let the kids enjoy the few years that are left to them,' Willy, one of the soldiers pleads, 'while they need still know nothing about it?' Is the classroom going to remain a torched battleground such as the one my students read about in 'All Quiet on the Western Front' — kids hunkering in the trenches of our schools while the adults fight over the eroded terrain of education? Will they become even more cut off from their own selves, just when they're getting to know who they are?
