AI's Evolution: From Chatbots to Ubiquitous Computing

Sora's AI Videos Arrive as Musk Pitches Image Tool for Making Memes

Member of the board, UCLA Daily Bruin Alumni Network; advisory board, Center for Ethical Leadership in the Media If OpenAI gets it right, Sora, the AI tool announced in February that's able to create photorealistic and high-quality video from simple text prompts, will upend the way video is created, with professional video creatives and Hollywood filmmakers among those expected to most feel its effect on their industry. Now those creatives -- and anyone with a $20 monthly subscription to ChatGPT Plus -- can test a version called Sora Turbo, which became available in the US last week (as part of OpenAI's 12 days of announcements). You can write a text prompt to create short clips, as well as upload photos and other videos as reference material for your prompts. The results are five to 20 seconds long, with resolutions between 480 and 1,080 pixels, and they can be delivered in wide-screen, vertical or square aspect ratios. There are also postgeneration editing features, like storyboard, remix and loop, for more fine-tuning, CNET's Katelyn Chedraoui notes. And though the demos of Sora over the past few months have been impressive -- with The Wall Street Journal noting that the AI videos "are good enough to freak us out" -- there are still some glitches to resolve, OpenAI acknowledged in a blog post. "It often generates unrealistic physics and struggles with complex actions over long durations," the post says. Think characters with extra arms and legs. The company is also limiting the testers who can create videos of humans, as it works to "address concerns about the misappropriation of likeness and deepfakes." In a nod to the ongoing issues about what training data is being used to feed AI engines (with some publishers alleging that OpenAI and others have co-opted their copyrighted material without permission), OpenAI wrote that Sora is trained on a "mix of publicly available data, proprietary data accessed through partnerships, and custom datasets developed in-house," including images supplied by humans such as employees. But there are already reports that OpenAI may have trained Sora on unlicensed video-game content (see TechCrunch and ExtremeTech). OpenAI didn't respond to my request for comment on the whole question of video-game content. The Sora news is a reminder that creating visuals with AI is going to be a big deal in the year ahead, with Google (Gemini image creator) and Meta (AI Studio) also creating image tools to get more users to engage with AI. Elon Musk's xAI last week also announced a photorealistic image editor, code-named Aurora, for its Grok chatbot. Musk, owner of social media platform X, touted Aurora as a way to "create awesome memes superfast." As for Sora, OpenAI also acknowledged the need for safety guardrails around the use of tools that may make it easier to create deepfakes -- which it wants others to help it solve for. "We're introducing our video generation technology now to give society time to explore its possibilities and co-develop norms and safeguards that ensure it's used responsibly as the field advances." the company wrote. Great, we're crowdsourcing AI safety parameters. I'm sure nothing can go wrong with that. Here are the other doings in AI worth your attention. Back in May, Google CEO Sundar Pichai debuted a bunch of AI tools, saying the company's vision included having its tech do the thinking for you -- specifically, by letting Google do all the "Googling for you" with features like AI Overviews. I bring this up because the company continues to deliver on that pitch and last week introduced a new version of its Gemini chatbot and a prototype of a tool called Project Mariner, under the blog headline "Our new AI model for the agentic era." What does that mean? "We have been investing in developing more agentic models, meaning they can understand more about the world around you, think multiple steps ahead, and take action on your behalf, with your supervision." Welcome to the new world of AI agents. That may include searching out a product for you and finding the best deal, scheduling meetings, interacting with spreadsheets to get answers to complex questions, and even playing games. This new generation of AI tech "will enable us to build new AI agents that bring us closer to our vision of a universal assistant," Pichai said in the blog post. "It can understand that it needs to press a button to make something happen," Dennis Hassabis, who oversees Google's DeepMind AI lab, said in an interview with The New York Times. "It can take action in the world." Mariner, developed as an extension for Google's Chrome browser, is designed to be used with "humans in the loop," company executives told the NYT. So, though it may fill your online shopping cart with groceries, you still need to press the buy button and make the purchase, they said. Still, the next generation of AI agents from Google, as well as from rivals including OpenAI and Anthropic, seem destined to bring us closer to a world where AI does more of the work for us. And in many cases, instead of us. If you think you've heard a lot about AI issues, challenges, innovations, products and services in 2024, that's nothing compared with how AI will continue to dominate the conversation in 2025. That's because more people are now trying out chatbots. OpenAI's No. 1 ranked ChatGPT chatbot saw its number of visitors double, to 3.9 billion, in November from the year before, according to Similarweb. And organizations, spending money on AI tools, continue to work on figuring out how to bring AI into the workplace pragmatically and ethically. So, I thought it was worthwhile to call out a few of the reports looking at the state of AI as we head into the New Year. Deloitte's survey of 1,874 early career versus longer-tenured workers highlighted their differing views on AI, with newer workers more excited about how AI could change the workforce for the better. "Their use of AI is such that one person we interviewed described AI as 'that first person you ask before going to a manager' for feedback and advice," the firm found. Deloitte's report is here. In its 2025 AI Business Predictions, PwC says most workforces will double in 2025 as they "welcome a host of new members to the team this year: digital workers known as AI agents." And while there's a lot of talk about which AI chatbots will win marketshare, PwC says having an AI strategy is more important for companies than picking the right large language model. "There will be many good options. Everyone will be using them. A shrewd strategy will instead emphasize what can set you apart" in using AI. When it comes to AI and higher education, Pearson found in its survey of more than 1,000 US college students that 58% said gen AI helped them get better grades. Meanwhile, 77% of nearly 3,500 faculty members in the US surveyed said they expect to use "AI to enhance their teaching methods." And when it comes to AI and the music industry, the International Confederation of Societies of Authors and Composers, which says it represents more than 5 million creators, conducted a global survey to determine how AI may affect the music industry in the not-too-distant future. It found that gen AI will "enrich tech companies while substantially jeopardising the income of human creators in the next five years." In a CNET survey, a quarter of smartphone owners said they don't find AI features helpful, 45% said they're reluctant to pay a monthly subscription fee for AI capabilities, and 34% said they have privacy concerns. Chatterbox Labs tested the leading large language models as part of its AI safety research, and after testing them across several categories of harm -- fraud; hate speech; illegal activity; misinformation; security and malware; self harm; sexually explicit content; and violence -- found that AI models from Anthropic and Amazon "showed the most progress in AI safety." Detailed test results can be found here. Local readers of the historic Ashland Daily Tidings in Oregon have been duped by scammers who used AI and, in some cases, stolen reporters' identities to produce fake news or AI slop for the newspaper, which shut down in 2023, according to an investigation by the nonprofit OPB media org. "Almost as soon as it closed, a website for the Tidings reemerged, boasting a team of eight reporters ... who cranked out densely reported stories every few days," it found. "The mysterious takeover of a more than 140-year-old news outlet offers a warning of ... what an online future supercharged by the next unregulated wave of technology from Silicon Valley companies may hold for news consumers." Though ChatGPT is positioning itself as a search engine working with publishers, a Tow Center report for the Columbia Journalism Review found that ChatGPT may actually misrepresent publishers' content. "While the company presents inclusion in its search as an opportunity to 'reach a broader audience,' a Tow Center analysis finds that publishers face the risk of their content being misattributed or misrepresented regardless of whether they allow OpenAI's crawlers." Asked for comment, OpenAI told the center that it's "collaborated with partners to improve in-line citation accuracy and respect publisher preferences."

OpenAI's plan to make ChatGPT the 'everything app' has never been more clear

OpenAI's quest to turn ChatGPT into the "everything app." Credit: Mashable composite: Zain bin Awais; Mike Coppola / Getty Images OpenAI is making major moves as it sets the table for 2025. To the casual observer, OpenAI has spent the past week flooding the tech world with announcements and rollouts -- not all of which gel into any sort of clear product agenda. It's all part of a holiday-timed event called "12 Days of OpenAI," a plan to ship a product or feature every day for 12 days marketed by the AI giant as bold and ambitious, even though some of the announcements, like a folder system for organizing your ChatGPT conversations, might seem like filler. But if you step back and look at the sum of its parts, you can see that OpenAI is deliberately laying a roadmap for ChatGPT to become the next "everything app." The term comes from Elon Musk when talking about transforming X, into "a single application that encompasses everything," as he told employees in a leaked 2023 all-hands meeting. "You can do payments, messages, video, calling, whatever you'd like, from one single, convenient place." Musk's plans to build X into an everything app have yet to come to fruition. But it's still a useful description for the strategy of embedding a technology so deeply into users' lives that they use it for all of their technological needs. OpenAI seems to be doing something similar with ChatGPT by launching user-friendly tools oriented around never having to leave the app. There's ChatGPT Search, its Google search engine competitor for accessing the web, Canvas, a kind of digital notepad for iterating on writing and code, Projects, a tool for creating and customizing projects that works with ChatGPT Search and Canvas, and Advanced Voice Mode, a vision and voice modalities tool that can see your screen and walk you through tasks. Beside you the whole time is ChatGPT, underpinned by GPT-4o, that can generate responses with text, audio, voice, and images. ChatGPT is on your iPhone now too, thanks to a partnership with Apple, the original lifestyle embedders. In the near future, OpenAI will take one step closer toward its goal of "achieving AGI" by putting all these tools together in an agent that can perform multi-step tasks on the user's behalf. This isn't just speculation. CEO Sam Altman said at OpenAI DevDay that "2025 is when agents will work," and CPO Kevin Weil reinforced this in a Reddit AMA, saying ChatGPT being able to perform tasks on its own will be "a big theme in 2025." Indeed, reports from Bloomberg and The Information say OpenAI's agentic tool, codenamed "Operator," will be able to book flights, write code, and generally browse the web. Meanwhile, AI model development might be running out of high quality training data and experiencing diminishing returns. And generative AI has a persistent hallucination problem that might not ever be fully resolved, which leads to the proliferation of AI slop, and also harmful misinformation, defamation, and potential copyright infringement. But despite genAI's proven unreliability for certain things and AI-weary public sentiment, OpenAI is doing its damndest to make the case for ChatGPT as the everything app with billion dollar stakes. Imagine you or your partner is pregnant for the first time and you have no idea what to do. You frantically use ChatGPT Search to research baby books, pregnancy diets and workouts, the best prenatal vitamins, birthing classes, why doctors hate doulas, baby gear, etc. You create a ChatGPT Project called "Baby Deliverables," and upload all of your research and notes, asking ChatGPT to help you plan a timeline of tasks before the baby's arrival and a budget for all the new expenses. As the pregnancy progresses, you use Advanced Voice Mode with vision to ask ChatGPT if the amount of swelling in you or your partner's feet is normal. You might even find yourself confiding in Advanced Voice Mode about the difficult hormonal changes or sharing ultrasound photos with the warm and friendly disembodied voice who's never busy, unlike family and friends. Soon, you might even be able to compile a list of gifts and automatically populate a baby shower registry, or ask ChatGPT to research the best strollers and buy the best-rated one within your price range. All of this is already or likely soon to be possible within ChatGPT. And that's exactly what OpenAI wants -- particularly since this scenario seeps beyond the 9-to-5 work productivity use case and into everyday companionship. The everything app concept isn't something Musk came up with. In fact, he was referencing apps in China like WeChat, a messaging, calling, social media, news, and payment app all rolled into one. But it was really Apple that championed the concept with the "walled garden" ecosystem. Apple is famous for slick, beautifully-designed products that seamlessly integrate with other Apple devices, while being incompatible with non-Apple products. Apple sells its users on the promise of realizing their untapped potential, or as Mashable's Chris Taylor calls it, "aspirational creativity," thus locking them into a neverending product cycle. With its reality distortion field, Apple's best trick is to simultaneously make consumers feel like they need Apple devices to be the most productive and creative versions of themselves while defending its anti-competitive business practices. As Cory Doctorow puts it in his description of the cult of mac, "Apple's most valuable intangible asset isn't its patents or copyrights -- it's an army of people who believe that using products from a $2.89 trillion multinational makes them members of an oppressed religious minority whose identity is coterminal with the interests of Apple's shareholders." When you look at OpenAI's recent launches, it seems like Sam Altman has taken a leaf out of the Steve Jobs playbook. "We really want to make ChatGPT as frictionless and easy to use everywhere," said Altman in a meta moment announcing the ChatGPT and iOS integration. "We love Apple devices and so this integration is one that we're very very proud of." ChatGPT's new features and updates seem to be carefully designed with the consumer in mind: paying special attention to the needs of a user's workflow or overall online behavior. In other words, the promise of enhanced productivity and creative fulfillment is just a monthly payment away. ChatGPT subscriptions are a critical part of OpenAI's revenue model. According to financial documents reviewed by the New York Times, around 10 million users pay $20 for a ChatGPT Plus subscription and OpenAI plans to raise the price to $44 a month over the next five years. And that was reported before OpenAI announced ChatGPT Pro for $200 a month. Combine that with OpenAI's revenue projections of $100 billion by 2029 and you can start to see the company's plan from novelty chatbot to everything app unfold. That all hinges on making ChatGPT more valuable to its users. ChatGPT has already started to move away from the restricted chatbot experience. In a few short years, it's evolved to support image, audio, and video in addition to text and will soon have more autonomous abilities. Speaking at Stanford University last spring, Altman called GPT-4 "the dumbest model any of you will ever have to use again," and described the forthcoming GPT-5 as a "significant leap forward" during an interview at the Aspen Ideas Festival. That's assuming the much-hyped GPT-5 is still part of the plan. In the meantime, OpenAI is rounding out ChatGPT with updates and features that emphasize its everyday usefulness. But just as OpenAI's allegedly altruistic mission "to ensure that AGI... benefits all of humanity," merits new scrutiny in the company's attempt to become for-profit, so does the intention behind its flurry of product announcements. OpenAI is a business and the goal of any business is to sell you stuff. Some of the new features may seem minor, but it's all part of a broader plan to hit that revenue projection. And maybe achieve AGI along the way.

The most popular AI tools of 2024 (and what that even means)

Popularity in the tech world is hard to measure. I've talked at length about this in my discussions of programming language popularity. It really comes down to what you use to measure popularity -- and how available those metrics are to those doing the analysis. When it comes to AI, that's even more of a challenge because AI can be a stand-alone tool or can be embedded deeply in other products. For example, the generative fill tool in Photoshop is very popular (and I quickly found it to be indispensable), but it won't show up in any public AI tool metrics. Also: I'm an AI tools expert, and these are the only two I pay for Exploding Topics is a company that analyzes trends based on web searches, conversations and mentions. It recently took on the challenge of determining AI tool popularity. Its approach was to collect data from two reliable web analytics platforms, Semrush and Similarweb, and calculate total traffic volume for 20 tools. The data is from August and is indicative of overall 2024 performance, especially since ChatGPT was and remains such a strongly dominant market leader. Let's look at the top ten performance leaders by market share. As the chart above shows, share numbers dropped rapidly once we are out of the top five. ChatGPT is, of course, the OG of generative AI services. In many ways, it's still the best, with enormous investment behind it, and an ever-growing war chest from its sales of services. Just recently, it announced its text-to-video application, Sora, and a $200/mo service for those who just can't get enough of that tasty AI goodness. Canva, a popular web-based image editing and marketing tool, has gone big into AI features, including a tool that automatically generates templates, background removal, automatic resizing to fit various social media formats, content suggestions, and text-to-image. DeepL is an online translation service that translates 33 languages, offers AI-powered edits, and can translate files including PDFs, Word documents, and PowerPoint decks. The company pitches its services as a "Language AI platform" that can integrate into a wide range of applications. Google Gemini has just received a whole slew of upgrades as part of the Gemini 2.0 introduction last week. While the product hasn't performed all that well in my programming tests, Google and Alphabet CEO Sundar Pichai reports that, "Our Al Overviews now reach 1 billion people." Those metrics may not be measurable by Semrush and Similarweb, which may be why its share numbers are lower than you'd expect given Pichai's boast. Character.AI allows users to create customized chatbots that simulate certain personalities. Because people are people, Character.AI has become quite controversial for what it allows, and for what people have tried to do to with their virtual characters. So, you know, if the AIs do decide to take over the world and destroy humanity, they'll be able to cite this site for justification. Remove.bg removes backgrounds from pictures. Sure, iOS and MacOS do that. Photoshop does that. Many apps do that. But hey, here's a simple web app that does fairly high-quality background removal, blurs backgrounds, and lets you add new backgrounds. It's a fee-based service, and since so many basic OS features are now beginning to add background removal, we're guessing that market share will probably drop over the next year. JanitorAI (I can't fully get over the name) is another AI character generator, but this one focuses more on online role playing than on deeper character characterizations. You can't see much from the site without logging in, but from the recent activity page that's publicly presented, it appears many of the users are creating anime-style characters. QuillBot bills itself as an online writing assistant, with a variety of tools including a grammar checker, plagiarism checker, AI detector, paraphraser, summarizer, citation generator, translator, and a tool called "Flow" that's essentially an AI-equipped word processor. I tested QuillBot in my AI detector shootout, and it got an 80% correct score. Grammarly existed long before the current generative AI boom. It was launched in 2009 in Kyiv, Ukraine, as a subscription service for students to help with grammar and spelling. Today, Grammarly is a fully-fleshed-out writing assistant that checks and suggests spelling, grammar, tone, and style and warns of possible plagiarism. We tested Grammarly's beta AI detector, and it didn't do all that well, scoring just a 40% accuracy result. Anthropic, the company behind the Claude chatbot, has been making a bunch of very big deals. It boasts a $4 billion investment from Amazon (with the intent that Claude will power "remarkable Alexa"), a $2 billion investment from Google, and an integration with Slack. That said, the 3.5 Sonnet version of Claude didn't do all that well in my programming tests. All of the services in our next batch have market shares under 1.5%. You may find some unexpected names there (or not find some you may expect). I'll discuss my observations after we get through this lightning round of the Under 1.5% Club. ChatGPT as the clear winner is completely expected. But what this list reminds us about is that AIs aren't just chatbots. Of the top five, only two are what we would now call "traditional" prompt-to-text chatbots. The other three perform translation, graphics support, and create characters. Also: Google's AI podcast tool transforms your text into stunningly lifelike audio - for free To some degree, I was surprised to see Microsoft's Copilot down in the Under 1.5% Club. That said, Copilot now permeates much of Microsoft's offerings and it's hard to pin it down as one monolithic beast for the purpose of measurement. That said, Bing isn't Google and Copilot isn't ChatGPT (or Gemini, for that matter). Some services I've spent a lot of time with, like Midjourney and Notion's AI, didn't show up at all. Midjourney spent most of its time as a Discord channel and only recently had a web interface. Tools like Notion AI live inside other products and may not stand alone enough to be visible to a survey like this. Way, way back, in the most ancient of times (1986), I wrote an article entitled "Artificial Intelligence as a systems component," for the long-defunct Computer Design magazine. At that time, my thesis was that AI would find its way inside many applications, to the point where it would just be a component of a much larger offering. Also: 25 AI tips to boost your programming productivity with ChatGPT To an extent, I was right. Copilot's features and Notion AI are just add-ons to another marquee offering. What I didn't realize at the time (because I was AI decades before AI was cool) was that AI would become a high-value marketing buzzword. Of course, back then, AI was a lot of work, showed only limited value, and wasn't available to muggles. Now, with ChatGPT and other offerings, it's become very mainstream and it's widely apparent that AI will change pretty much everything. So, as we transition from 2024 to 2025, expect AI to become even more of a hot topic. If 2023 was the year of "gee, check out what this can do" and 2024 the year of "let me get some of that," 2025 will be the year when solutions solidify, and AI gets more integrated into business and personal activities. In 2025, expect to see some fallout both from the inevitable chilling effects of this technology and from companies who over-reached, over-bragged, over-promised, and under-delivered. Also: AI is moving undercover at work in 2025, according to Deloitte's Tech Trends report Also expect to see ChatGPT continue to get stronger, but so too will Gemini and Copilot. Neither Google nor Microsoft are accustomed to or willing to be upstaged by rivals, so we can expect their offerings to become more of the solid, reliable solutions both companies are known for. As for Apple, Amazon, and Meta, the jury is still out. Meta sure seems to get it, while Alexa hasn't changed all that much in terms of intelligence since its inception, and Siri is still as stubbornly unhelpful as it was when it debuted. Will these three step more strongly into the AI spotlight with game-changing offerings? What do you think? Let us know in the comments below.

Is Satya Nadella's 2014 Vision Finally Becoming a Reality? | AI Origins & Evolution

'Ambient intelligence and ubiquitous computing' isn't free from challenges, though. In the past two decades, Microsoft has successfully cracked the code for personal computing, long striving to go above and beyond. Their strides in AI have been quite dramatic over the past year or so. In a podcast, CEO Satya Nadella reminded the public that he envisioned Microsoft being at the forefront of AI nearly a decade ago, in 2014. "I am here for the same reason I think most people join Microsoft - to change the world through technology that empowers people to do amazing things," he said. When appointed CEO, he wrote in an internal memo: "I believe that over the next decade, computing will become even more ubiquitous, and intelligence will become ambient." Nadella reminisced about Microsoft's attempts to bring the best products to the market, eventually getting the browser game right with Edge and winning the cloud market, albeit ceding ground in mobile and search. With AI, it is too early to draw a conclusion, but Microsoft has indeed made it big. In a memo to the committee that picked the CEO, Nadella again used the phrase "ambient intelligence and ubiquitous computing". Later, he simplified it to say: "mobile-first, and cloud later". "My PR folks came and said, 'What the heck is this? Nobody will understand this'," laughed Nadella. Ten years down the line, Nadella's once perplexing phrase is now our reality. Today, some of Microsoft's products align with 'ubiquitous and ambient computing', notably its AI PCs and Copilot Vision. Qualcomm, AMD, and Intel are going all in with Microsoft to bring AI features to PCs using the Copilot + program. In this regard, former Intel CEO Pat Gelsinger said, "When I think about the PC market, this is the most exciting moment in 25 years since the arrival of WiFi." Likewise, Qualcomm CEO Cristiano Amon called it the "most important development" since the Microsoft 95 operating system. In fact, Microsoft's recent Phi-4 small language models pack in a lot of power with just 14 billion parameters. AIM spoke to Harkirat Behl, one of the creators of the Phi series of models, who said that once the architecture is released on Hugging Face, developers will be able to quantise it and optimise it to work locally on laptops and PCs. But Microsoft's leaps were not devoid of challenges. For instance, the company began rolling out the preview version of Microsoft Recall after it was halted in June due to privacy loopholes. The feature takes screenshots of your activity and uses AI to help you recall and find content. However, recent reports suggest that the tool captures screenshots of payment info and personal identification numbers, even when certain filters are enabled. "If a bad actor has access to your computer and knows your PIN, they could view Recall bypassing the biometric security checks. They don't even need physical access to the PC." "I was able to access the Recall app and view the timeline on a remote computer by using TeamViewer, a popular remote access application," said Avram Piltch from Tom's Hardware. However, AI PCs do seem promising for on-device AI capabilities. Intel surveyed 6,000 people in European markets, such as Germany, France, and the UK. About 44% believed that AI PCs were a gimmick, and a whopping 86% were "concerned about the privacy and security of their data when using an AI PC". During the Copilot + launch event in May, Nadella said that he expects 50 million sales of AI PCs in the next twelve months. However, according to a Gartner report, global PC sales have only declined in the past few months. "Even with a full lineup of Windows-based AI PCs for both Arm and x86 in the third quarter of 2024, AI PCs did not boost the demand for PCs since buyers have yet to see their clear benefits or business value," said Mikako Kitagawa, director analyst at Gartner. Nadella also recalled the time OpanAI CEO Sam Altman sought help for the company with Azure credits. This was after OpenAI went to Google Cloud Platform (GCP), but couldn't find a solution and eventually came to Microsoft. It was at this moment that Altman and Co. spoke to Nadella about Transformers and natural language. "You needed some breakthrough - and maybe the way to do it was how we schematise the human brain and what it does through language and reasoning," he said. However, the industry and the government don't seem to be fans of the Microsoft-OpenAI bond. Recently, Alphabet, which owns Google, asked the Federal Trade Commission to end their partnership. Reports suggest that companies that buy OpenAI's technology may have to pay an additional cost if they do not use Microsoft's servers to run it. Once a founder of OpenAI, Elon Musk, has had a long tussle with the company since his departure. In his lawsuits that claim OpenAI has abandoned its once set 'not for profit' mission, he has now dragged Micorosft, accusing them of anti-competitive and monopolistic practices. Microsoft, on the other hand, has maintained its stance -- it will do everything it takes to get OpenAI on top. A few months ago, the former also participated in the $6.6 billion funding for OpenAI. When OpenAI announced updates to ChatGPT search and how it seamlessly integrates with the iPhone, Nadella may have felt like a proud dad, as OpenAI was achieving what Microsoft could not. "I've been trying to get an Apple search deal for about 10 years, and so when Tim finally did a deal with Sam, I was the most thrilled person," he said. Moreover, OpenAI may also remove a clause from its agreement with Microsoft which disables its access to OpenAI's advanced models when they reach 'AGI'. It was included to prevent a large corporation like Microsoft from gaining access to powerful AI systems and prevent any misuse through commercial applications. Removing the clause would further tempt Microsoft to pour capital into OpenAI. But when will OpenAI achieve AGI? Is it going to be after the ongoing '12 days of OpenAI'? The world waits in anticipation.

AI is moving undercover at work in 2025, according to Deloitte's Tech Trends report

The 16th annual Deloitte Tech Trends Report analyzes the many changes across the AI landscape during the past year. Every year, Deloitte releases its Tech Trends report, which takes a deep dive into the past year's technological landscape and identifies macro industry trends that will play a pivotal role in digital transformation in the upcoming 18 to 24 months. Unsurprisingly, artificial intelligence (AI) was a key area in this year's edition, released today -- just not in the same way as in the past. Even though AI is just as buzzy this year as it was when its popularity exploded two years ago, there have been some significant industry shifts in how technology is perceived, adapted, and deployed by consumers and organizations, which, according to the report, "is being woven into the fabric of our lives." Also: OpenAI rolls out Canvas to all ChatGPT users - and it's a powerful productivity tool Mike Bechtel, Deloitte's chief futurist and one of the report's authors, explained that, much like electricity or the World Wide Web, technologies people rarely think about daily yet rely on for their roles, AI is becoming an underlying layer for key business operations. "The difference between the next 18 to 24 months and the last 18 to 24 months is that AI is moving undercover," said Bechtel. "What we're seeing is that it's sort of melting into becoming the foundation or sub-structure of all the other business-oriented things that we need to do and think about." As AI permeates nearly all business operations, the initial apprehension about the technology is fading. Instead of wondering if they should go all-in on AI, business leaders now face the new hurdle of determining how to exploit the technology, exploring concepts like upgrading hardware, small language models (SLMs), agentic AI, and more. When AI chatbots like ChatGPT took the world by storm, the underlying large language models (LLMs) became sought after to optimize business operations. In these case scenarios, a business would have one chatbot that could do it all. "For the last two years, it felt like AI was a monolithic thing -- one chat window for all of our needs and all of our curiosities," said Bechtel. "What we're starting to see is this fractal explosion where it's going to be AIs (plural), dozens and then hundreds and eventually, thousands of domain-specific agents trained on domain-specific data." Furthermore, the report explains that, due to the flexible nature of SLMs, each assistant can help organizations carry out a specific task, such as applying for a grant, delivering a financial report, and summarizing an inspection report, all mundane tasks in which training an LLM would be inefficient. Also: Reddit's latest AI update makes finding the answers you want much easier These SLMs will work together, creating a team of virtual assistants that organizations can depend on to complete many tasks. Organizations may approach AI similarly to apps. 'There's an app for that' could well become 'There's an agent for that.' "It's less of a single super suit; it's more like a posse, an Avengers team, if you will, of specially powered agents with specialized superpowers that you call on for special needs," said Bechtel. Other benefits of SLMs are they can be run on-device, are more cost-efficient, require less data, and are often open-sourced. During the past year, there has been an explosion in the amount of AI-centric hardware equipped with more advanced computational power. This hardware can run AI applications on-device and incorporate AI features and workflows. "An enterprise and employee physical computer refresh is about to start, the likes of which we haven't seen in 15 years," said Bechtel. "For the first time in a generation, physical kit, processors, servers, networking, laptops, they can be the key between getting to the future you want, and being stuck in the past." Also: Reddit's latest AI update makes finding the answers you want much easier This development is forcing companies to consider upgrading their employees' physical devices, as the AI transformation they envision for their companies is impossible without the proper hardware to support it. This resurgence in prioritizing hardware investments to meet digital transformation goals led the report to dub one of the six trends: "Hardware is eating the world." The most obvious example of this trend is Nvidia's huge growth in popularity during the past year, with its GPUs becoming one of the hottest commodities, and placing the company as one of the world's most valuable firms. Deloitte projects the global AI chip market will grow from $50 billion in 2024 to $110bn in 2027 in a conservative forecast or up to $400bn in 2027 in an optimistic forecast. Ultimately, when deciding when to invest resources into upgrading an organization's hardware, Bechtel said his go-to advice is to "lead with need." "If you have a set of strategic capabilities that you're champing at the bit to deliver and your hardware estate is your limiting reagent, then, by all means, it's time," said Bechtel. "If you're window shopping sparkly hardware, just because it's sparkly, that might not be the most value-added decision." Although the overall AI landscape has significantly changed from 2023, the advice on what area businesses should start with is the same as last year: data. Ultimately, whether your company is interested in using an LLM or SLM, you will need clean, organized, and up-to-date data to get the right results. "I think recommendation number one that I give to all my clients in every sector in every geography is clean and normalize and govern your individual data sets before you think of dumping them into an AI," said Bechtel. About three-quarters of the clients he speaks to go into meetings thinking they want an AI project, but they walk out realizing they need a data management and governance project. Also: How Cerebras boosted Meta's Llama to 'frontier model' performance To avoid "garbage in, garbage squared," the report suggests data-labeling costs are a big driver of AI investment. Furthermore, Deloitte's 2024 State of Generative AI in the Enterprise Q3 report found that 75% of surveyed organizations increased their investments in data cycle management because of generative AI. "When you look at the word 'IT' right, you tend to think of the T right, like, 'oh, technology shiny,' but the information piece is easily half of the game," said Bechtel.

AI remains a top concern for enterprises, but for different reasons than last year

The 16th annual Deloitte Tech Trends Report analyzes the many changes across the AI landscape during the past year. Every year, Deloitte releases its Tech Trends report, which takes a deep dive into the past year's technological landscape and identifies macro industry trends that will play a pivotal role in digital transformation in the upcoming 18 to 24 months. Unsurprisingly, artificial intelligence (AI) was a key area in this year's edition, released today -- just not in the same way as in the past. Even though AI is just as buzzy this year as it was when its popularity exploded two years ago, there have been some significant industry shifts in how technology is perceived, adapted, and deployed by consumers and organizations, which, according to the report, "is being woven into the fabric of our lives." Also: OpenAI rolls out Canvas to all ChatGPT users - and it's a powerful productivity tool Mike Bechtel, Deloitte's chief futurist and one of the report's authors, explained that, much like electricity or the World Wide Web, technologies people rarely think about daily yet rely on for their roles, AI is becoming an underlying layer for key business operations. "The difference between the next 18 to 24 months and the last 18 to 24 months is that AI is moving undercover," said Bechtel. "What we're seeing is that it's sort of melting into becoming the foundation or sub-structure of all the other business-oriented things that we need to do and think about." As AI permeates nearly all business operations, the initial apprehension about the technology is fading. Instead of wondering if they should go all-in on AI, business leaders now face the new hurdle of determining how to exploit the technology, exploring concepts like upgrading hardware, small language models (SLMs), agentic AI, and more. When AI chatbots like ChatGPT took the world by storm, the underlying large language models (LLMs) became sought after to optimize business operations. In these case scenarios, a business would have one chatbot that could do it all. "For the last two years, it felt like AI was a monolithic thing -- one chat window for all of our needs and all of our curiosities," said Bechtel. "What we're starting to see is this fractal explosion where it's going to be AIs (plural), dozens and then hundreds and eventually, thousands of domain-specific agents trained on domain-specific data." Furthermore, the report explains that, due to the flexible nature of SLMs, each assistant can help organizations carry out a specific task, such as applying for a grant, delivering a financial report, and summarizing an inspection report, all mundane tasks in which training an LLM would be inefficient. Also: Reddit's latest AI update makes finding the answers you want much easier These SLMs will work together, creating a team of virtual assistants that organizations can depend on to complete many tasks. Organizations may approach AI similarly to apps. 'There's an app for that' could well become 'There's an agent for that.' "It's less of a single super suit; it's more like a posse, an Avengers team, if you will, of specially powered agents with specialized superpowers that you call on for special needs," said Bechtel. Other benefits of SLMs are they can be run on-device, are more cost-efficient, require less data, and are often open-sourced. During the past year, there has been an explosion in the amount of AI-centric hardware equipped with more advanced computational power. This hardware can run AI applications on-device and incorporate AI features and workflows. "An enterprise and employee physical computer refresh is about to start, the likes of which we haven't seen in 15 years," said Bechtel. "For the first time in a generation, physical kit, processors, servers, networking, laptops, they can be the key between getting to the future you want, and being stuck in the past." Also: Reddit's latest AI update makes finding the answers you want much easier This development is forcing companies to consider upgrading their employees' physical devices, as the AI transformation they envision for their companies is impossible without the proper hardware to support it. This resurgence in prioritizing hardware investments to meet digital transformation goals led the report to dub one of the six trends: "Hardware is eating the world." The most obvious example of this trend is Nvidia's huge growth in popularity during the past year, with its GPUs becoming one of the hottest commodities, and placing the company as one of the world's most valuable firms. Deloitte projects the global AI chip market will grow from $50 billion in 2024 to $110bn in 2027 in a conservative forecast or up to $400bn in 2027 in an optimistic forecast. Ultimately, when deciding when to invest resources into upgrading an organization's hardware, Bechtel said his go-to advice is to "lead with need." "If you have a set of strategic capabilities that you're champing at the bit to deliver and your hardware estate is your limiting reagent, then, by all means, it's time," said Bechtel. "If you're window shopping sparkly hardware, just because it's sparkly, that might not be the most value-added decision." Although the overall AI landscape has significantly changed from 2023, the advice on what area businesses should start with is the same as last year: data. Ultimately, whether your company is interested in using an LLM or SLM, you will need clean, organized, and up-to-date data to get the right results. "I think recommendation number one that I give to all my clients in every sector in every geography is clean and normalize and govern your individual data sets before you think of dumping them into an AI," said Bechtel. About three-quarters of the clients he speaks to go into meetings thinking they want an AI project, but they walk out realizing they need a data management and governance project. Also: How Cerebras boosted Meta's Llama to 'frontier model' performance To avoid "garbage in, garbage squared," the report suggests data-labeling costs are a big driver of AI investment. Furthermore, Deloitte's 2024 State of Generative AI in the Enterprise Q3 report found that 75% of surveyed organizations increased their investments in data cycle management because of generative AI. "When you look at the word 'IT' right, you tend to think of the T right, like, 'oh, technology shiny,' but the information piece is easily half of the game," said Bechtel.