{"id":132048,"date":"2026-01-13T20:08:09","date_gmt":"2026-01-13T20:08:09","guid":{"rendered":"https:\/\/www.newsbeep.com\/us-ca\/132048\/"},"modified":"2026-01-13T20:08:09","modified_gmt":"2026-01-13T20:08:09","slug":"11-things-uc-berkeley-ai-experts-are-watching-for-in-2026","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/us-ca\/132048\/","title":{"rendered":"11 things UC Berkeley AI experts are watching for in 2026"},"content":{"rendered":"<p>In just a few years, AI has gone from a tech novelty to a society-altering force. It has <a href=\"https:\/\/www.chemistryworld.com\/news\/chemistry-will-no-longer-be-an-exclusive-club-how-ai-is-changing-omar-yaghis-work\/4020898.article\" rel=\"nofollow noopener\" target=\"_blank\">revolutionized scientific research<\/a>, <a href=\"https:\/\/www.chronicle.com\/article\/ai-has-joined-the-faculty\" rel=\"nofollow noopener\" target=\"_blank\">classrooms<\/a> and <a href=\"https:\/\/www.mckinsey.com\/capabilities\/tech-and-ai\/our-insights\/superagency-in-the-workplace-empowering-people-to-unlock-ais-full-potential-at-work\" rel=\"nofollow noopener\" target=\"_blank\">workplaces<\/a>. People around the planet have turned to it to <a href=\"https:\/\/www.theguardian.com\/business\/2025\/oct\/27\/ai-authors-writers-block-bloomsbury-chief-book-publisher-shares\" rel=\"nofollow noopener\" target=\"_blank\">break through writer\u2019s block<\/a> and to <a href=\"https:\/\/www.fastcompany.com\/90898619\/i-asked-chatgpt-to-contest-my-parking-ticket-what-followed-was-a-thing-of-beauty\" rel=\"nofollow noopener\" target=\"_blank\">contest parking tickets<\/a>.\u00a0<\/p>\n<p>But that growth has not come without risks. The <a href=\"https:\/\/www.nytimes.com\/2025\/10\/27\/briefing\/is-ai-a-bubble.html\" rel=\"nofollow noopener\" target=\"_blank\">economy<\/a> is increasingly propped up by bets about the technology\u2019s future \u2014 <a href=\"https:\/\/www.ft.com\/content\/6cc87bd9-cb2f-4f82-99c5-c38748986a2e\" rel=\"nofollow noopener\" target=\"_blank\">80% of U.S. stock gains<\/a> last year came from AI companies. <a href=\"https:\/\/calmatters.org\/economy\/technology\/2025\/12\/california-ai-regulation-targeted-in-trump-order\/\" rel=\"nofollow noopener\" target=\"_blank\">Policy battles<\/a> are ramping up as some seek to rein in tech companies and others opt for an unregulated Wild West. <a href=\"https:\/\/www.washingtonpost.com\/technology\/2026\/01\/06\/x-grok-deepfake-sexual-abuse\/\" rel=\"nofollow noopener\" target=\"_blank\">Deepfake and explicit<\/a> videos are causing harm and blurring what is real in an already fragmented information environment.\u00a0<\/p>\n<p>As a global leader in the development of AI technology as well as research into the ethics, policies and practices around its use, UC Berkeley\u2019s experts are at the forefront of this rapidly changing technology.\u00a0Below, UC Berkeley News asked some of the campus\u2019s leading AI experts to summarize in 100 words \u2014 and in a short phrase \u2014 the developments they\u2019ll be monitoring in 2026.\u00a0<\/p>\n<p>\u2018Will the AI bubble burst?\u2019<\/p>\n<p><img fetchpriority=\"high\" decoding=\"async\" width=\"455\" height=\"441\" data-ccwcag-attachment-id=\"125796\" data-ccwcag-attachment=\"{\" disable_page_edit=\"\" src=\"https:\/\/www.newsbeep.com\/us-ca\/wp-content\/uploads\/2026\/01\/StuartRussell_cropped.jpg\" alt=\"Stuart Russell profile\" class=\"wp-image-125796\" style=\"width:315px;height:auto\"\/>Stuart Russell<\/p>\n<p>Current and planned <a href=\"https:\/\/www.theguardian.com\/technology\/2025\/nov\/02\/global-datacentre-boom-investment-debt\" rel=\"nofollow noopener\" target=\"_blank\">spending on data centers<\/a> represents the <a href=\"https:\/\/www.reuters.com\/graphics\/USA-ECONOMY\/AI-INVESTMENT\/gkvlqbgxkpb\/\" rel=\"nofollow noopener\" target=\"_blank\">largest technology project in history<\/a>. Yet many observers describe a bubble that is about to burst: revenues are underwhelming, the performance of large language models seems to have plateaued, and there are clear theoretical limits on their ability to learn straightforward concepts efficiently.\u00a0<\/p>\n<p>If the bubble bursts, the economic damage will be severe. But for the bubble not to burst, breakthroughs will need to happen that take us close to artificial general intelligence. AI developers have no cogent proposal for how to control such systems, leading to risks far greater than economic damage.<\/p>\n<p>\u2014 <a href=\"https:\/\/vcresearch.berkeley.edu\/faculty\/stuart-russell\" rel=\"nofollow noopener\" target=\"_blank\">Stuart Russell<\/a>, professor of electrical engineering and computer sciences<\/p>\n<p>\u2018Can we trust anything anymore?\u2019<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" width=\"670\" height=\"729\" data-ccwcag-attachment-id=\"125791\" data-ccwcag-attachment=\"{\" disable_page_edit=\"\" src=\"https:\/\/www.newsbeep.com\/us-ca\/wp-content\/uploads\/2026\/01\/HaneyFarid_ProfileCropped.jpg\" alt=\"Haney Farid profile\" class=\"wp-image-125791\" style=\"width:302px;height:auto\"\/>Hany Farid<\/p>\n<p>I will be watching the accelerating erosion of trust driven by increasingly convincing AI-generated media. In 2026, deepfakes will no longer be novel; they will be routine, scalable, and cheap, blurring the line between the real and the fake. This has profound implications for journalism, democracies, economies, courts and personal reputation.\u00a0<\/p>\n<p>I am especially concerned about the asymmetry: It takes little effort to create a fake, but enormous effort to debunk it after it spreads. How society adapts \u2014 technically, legally and culturally \u2014 to a world where seeing is no longer believing will be critical.<\/p>\n<p>\u2014 <a href=\"https:\/\/vcresearch.berkeley.edu\/faculty\/hany-farid\" rel=\"nofollow noopener\" target=\"_blank\">Hany Farid<\/a>, professor of information<\/p>\n<p>\u2018AI-enabled discoveries that benefit people\u2019<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" width=\"1022\" height=\"1024\" data-ccwcag-attachment-id=\"137237\" data-ccwcag-attachment=\"{\" disable_page_edit=\"\" src=\"https:\/\/www.newsbeep.com\/us-ca\/wp-content\/uploads\/2026\/01\/jennifer-chayes-headshot-2024-1022x1024.jpg\" alt=\"Jennifer Chayes portrait\" class=\"wp-image-137237\" style=\"width:333px;height:auto\"  \/>Jennifer Chayes<\/p>\n<p>Major technology paradigm shifts like AI come with significant benefits and risks, and I expect AI will become ever more part of our daily lives in 2026.\u00a0<\/p>\n<p>Individuals and industries are finding exciting new uses for personalized agents and related technologies. For example, AI accelerates scientific discovery in ways that were previously unimaginable.\u00a0<\/p>\n<p>Conversations about the responsible and ethical use of AI should be prioritized across sectors and civil society. We must work collaboratively to mitigate AI\u2019s potential harms and find inclusive ways to empower people. <\/p>\n<p>Our challenge is to apply AI to advance knowledge, expand understanding and benefit humanity.<\/p>\n<p>\u2014\u00a0<a href=\"https:\/\/cdss.berkeley.edu\/people\/jennifer-chayes\" rel=\"nofollow noopener\" target=\"_blank\">Jennifer Chayes<\/a>, dean of the UC Berkeley College of Computing, Data Science, and Society<\/p>\n<p>\u2018Privacy risks created by chatbot logs\u2019<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" width=\"768\" height=\"1024\" data-ccwcag-attachment-id=\"137238\" data-ccwcag-attachment=\"{\" disable_page_edit=\"\" src=\"https:\/\/www.newsbeep.com\/us-ca\/wp-content\/uploads\/2026\/01\/Mulligan_profile-768x1024.jpg\" alt=\"Deirdre Mulligan profile\" class=\"wp-image-137238\" style=\"width:270px;height:auto\"  \/>Deirdre Mulligan<\/p>\n<p>People use AI chatbots for emotional support, spiritual guidance, relationship counseling, legal advice and intimacy, and they turn over reams of information about their follies, fantasies and fears.\u00a0<\/p>\n<p><a href=\"https:\/\/www.nytimes.com\/2025\/08\/26\/technology\/chatgpt-openai-suicide.html?searchResultPosition=2\" rel=\"nofollow noopener\" target=\"_blank\">Adam Raine\u2019s<\/a> ChatGPT logs <a href=\"https:\/\/www.documentcloud.org\/documents\/26078522-raine-vs-openai-complaint\/\" rel=\"nofollow noopener\" target=\"_blank\">cataloging struggles<\/a> before his death by suicide are central to his family\u2019s lawsuit. While Adam\u2019s chats are being used to address harms, users\u2019 logs risk disclosure in more troubling settings. A recent <a href=\"https:\/\/mashable.com\/article\/openai-court-ordered-chat-gpt-preservation-no-longer-required\" rel=\"nofollow noopener\" target=\"_blank\">court order <\/a>directed OpenAI to save all chats for a copyright lawsuit, and <a href=\"https:\/\/www.forbes.com\/sites\/thomasbrewster\/2025\/10\/20\/openai-ordered-to-unmask-writer-of-prompts\/\" rel=\"nofollow noopener\" target=\"_blank\">the Department of Homeland Security successfully demanded a user\u2019s prompts<\/a>.\u00a0<\/p>\n<p>Expect more demands on AI companies for personal data and lawsuits challenging how they collect and use it.<\/p>\n<p>\u2014 <a href=\"https:\/\/www.ischool.berkeley.edu\/people\/deirdre-mulligan\" rel=\"nofollow noopener\" target=\"_blank\">Deirdre Mulligan<\/a>, professor of information<\/p>\n<p>\u2018Relationship chatbots worsen human isolation\u2019<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" width=\"183\" height=\"275\" data-ccwcag-attachment-id=\"102802\" data-ccwcag-attachment=\"{\" disable_page_edit=\"\" src=\"https:\/\/www.newsbeep.com\/us-ca\/wp-content\/uploads\/2026\/01\/Halpern_BrittanyHoseaSmall_crop-183x275.jpg\" alt=\"A photo of Jodi Halpern outside in a garden\" class=\"wp-image-102802\" style=\"width:236px;height:auto\"  \/>Jodi Halpern<\/p>\n<p>This year will see the expansion of companion chatbots to young children. Already, one-third of teens prefer talking about their feelings with bots instead of people. Using the existing social media business model, companies manipulate and incentivize overuse. Sadly, teen dependence is associated with suicide. Yet \u201cbuddies\u201d for toddlers are an emerging market without guardrails.\u00a0<\/p>\n<p>In the short term, we need regulation until safety is established. Long-term, how will toddlers who learn \u201cempathy\u201d from sycophantic bots rather than people develop the mutual empathic curiosity crucial for successful relationships? How will they participate in democracy with people who have different perspectives?<\/p>\n<p>\u2014 <a href=\"https:\/\/publichealth.berkeley.edu\/people\/jodi-halpern\" rel=\"nofollow noopener\" target=\"_blank\">Jodi Halpern<\/a>, professor of public health\u00a0<\/p>\n<p>\u2018Whether robots can learn useful manipulation tasks\u2019<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" width=\"600\" height=\"603\" data-ccwcag-attachment-id=\"134936\" data-ccwcag-attachment=\"{\" disable_page_edit=\"\" src=\"https:\/\/www.newsbeep.com\/us-ca\/wp-content\/uploads\/2026\/01\/Headshot-KG-Large-1.jpg\" alt=\"A headshot of Ken Goldberg\" class=\"wp-image-134936\" style=\"width:317px;height:auto\"  \/>Ken Goldberg<\/p>\n<p>I\u2019m very concerned about widespread claims that humanoid robots will \u201csoon\u201d replace human workers. Significant advances are being made, but robots have nowhere near the dexterity of kitchen workers, car mechanics or construction workers.\u00a0\u00a0<\/p>\n<p>There is a <a href=\"https:\/\/news.berkeley.edu\/2025\/08\/27\/are-we-truly-on-the-verge-of-the-humanoid-robot-revolution\/\" rel=\"nofollow noopener\" target=\"_blank\">vast gap between the amount of data available<\/a> for training large language models such as ChatGPT and the amount of data available to train robots. Researchers around the world, including my own lab, are working on this, and I\u2019m watching for breakthroughs in how to close this data gap.<\/p>\n<p>\u2014 <a href=\"https:\/\/vcresearch.berkeley.edu\/faculty\/ken-goldberg\" rel=\"nofollow noopener\" target=\"_blank\">Ken Goldberg<\/a>, professor of engineering<\/p>\n<p>\u2018Progress on worker technology rights\u2019<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" width=\"560\" height=\"560\" data-ccwcag-attachment-id=\"125916\" data-ccwcag-attachment=\"{\" disable_page_edit=\"\" src=\"https:\/\/www.newsbeep.com\/us-ca\/wp-content\/uploads\/2026\/01\/Annette-Bernhardt-e1591154609301-560x560-1.webp.webp\" alt=\"Annette Bernhardt profile\" class=\"wp-image-125916\" style=\"width:309px;height:auto\"  \/>Annette Bernhardt<\/p>\n<p>In 2025, unions and other worker advocates began to develop a <a href=\"https:\/\/www.techpolicy.press\/evaluating-trends-and-challenges-in-state-regulation-of-workplace-technologies\/\" rel=\"nofollow noopener\" target=\"_blank\">portfolio of policies<\/a> to regulate employers\u2019 growing use of AI and other digital technologies \u2014 an area where workers currently have few rights.\u00a0<\/p>\n<p>This year I will be watching for progress in legislation to establish guardrails around electronic monitoring and algorithmic management, addressing harms such as automated firing, discrimination, invasive surveillance and profiling of union organizers. <\/p>\n<p>Equally important will be the success of bills requiring that humans have a meaningful role in critical decisions that affect our lives (like health care), which is protective of both workers and the public.<\/p>\n<p>\u2014 <a href=\"https:\/\/laborcenter.berkeley.edu\/people\/annette-bernhardt\/\" rel=\"nofollow noopener\" target=\"_blank\">Annette Bernhardt<\/a>, director of the UC Berkeley Labor Center\u2019s <a href=\"https:\/\/laborcenter.berkeley.edu\/technology-and-work\/\" rel=\"nofollow noopener\" target=\"_blank\">Technology and Work Program<\/a><\/p>\n<p>\u2018Weaponization of AI against workers\u2019<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" width=\"849\" height=\"1024\" data-ccwcag-attachment-id=\"124454\" data-ccwcag-attachment=\"{\" disable_page_edit=\"\" src=\"https:\/\/www.newsbeep.com\/us-ca\/wp-content\/uploads\/2026\/01\/IMG_6049-849x1024.jpg\" alt=\"Nicole Holliday portrait\" class=\"wp-image-124454\" style=\"width:281px;height:auto\"  \/>Nicole Holliday<\/p>\n<p>My <a href=\"https:\/\/pubs.aip.org\/asa\/jasa\/article\/158\/6\/4361\/3373813\/Socially-prescriptive-speech-technologies\" rel=\"nofollow noopener\" target=\"_blank\">research<\/a> focuses on AI that aims to evaluate how people speak. Programs like the Zoom Revenue Accelerator and Read AI are being used by companies, often without worker knowledge or consent, to rate employees.\u00a0<\/p>\n<p>These systems rate speech on parameters like \u201ccharisma,\u201d but because they are black-box algorithms, no one knows how they work. They are trained on \u201cidealized\u201d speech, so they show systematic bias against neurodivergent speakers, second-language English speakers, and anyone who speaks a stigmatized dialect. I\u2019m watching to see how systems are weaponized against workers and how the courts deal with these patterns of bias.\u00a0<\/p>\n<p>\u2014 <a href=\"https:\/\/ls.berkeley.edu\/people\/nicole-holliday\" rel=\"nofollow noopener\" target=\"_blank\">Nicole Holliday<\/a>, associate professor of linguistics<\/p>\n<p>\u2018AI\u2019s effect on political conflict\u2019<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" width=\"1015\" height=\"1024\" data-ccwcag-attachment-id=\"125464\" data-ccwcag-attachment=\"{\" disable_page_edit=\"\" src=\"https:\/\/www.newsbeep.com\/us-ca\/wp-content\/uploads\/2026\/01\/Jonathan_Stray_2023_edit-scaled-1-1015x1024.gif\" alt=\"Jonathan Stray portrait\" class=\"wp-image-125464\" style=\"width:299px;height:auto\"  \/>Jonathan Stray<\/p>\n<p>The lack of good definitions and evaluations of \u201cpolitically neutral AI\u201d is a problem for democracy. The White House recently issued an executive order requiring \u201cpolitically neutral and unbiased\u201d AI for federal contractors. They didn\u2019t provide a definition.\u00a0<\/p>\n<p>We recently saw Grok turn into \u201cMechaHitler.\u201d Next time the change could be more subtle and we might never detect the bias. With any luck this year will see advances in defining and implementing \u201cpolitically neutral\u201d AI. What does this idea even mean? Should an AI ever persuade you? And how would we build models that have this property?<\/p>\n<p>\u2014 <a href=\"https:\/\/humancompatible.ai\/people#jonathan-stray\" rel=\"nofollow noopener\" target=\"_blank\">Jonathan Stray<\/a>, senior scientist at the <a href=\"https:\/\/humancompatible.ai\/\" rel=\"nofollow noopener\" target=\"_blank\">UC Berkeley Center for Human-Compatible AI<\/a><\/p>\n<p>\u2018More sophisticated deepfakes\u2019<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" width=\"885\" height=\"1024\" data-ccwcag-attachment-id=\"125860\" data-ccwcag-attachment=\"{\" disable_page_edit=\"\" src=\"https:\/\/www.newsbeep.com\/us-ca\/wp-content\/uploads\/2026\/01\/Crittenden_cropped-885x1024.jpg\" alt=\"Camille Crittenden profile\" class=\"wp-image-125860\" style=\"width:282px;height:auto\"  \/>Camille Crittenden<\/p>\n<p>This year will mark when video and audio manipulation goes mainstream. Deepfakes have proliferated across social media \u2014 from whimsical images of beavers piloting planes or the pope in a puffy coat to harmful uses, including child sexual abuse material, political disinformation and financial fraud. <\/p>\n<p>What changes in 2026 is scale: Powerful tools and platforms are making sophisticated audio and video manipulation cheap, fast and accessible. <a href=\"https:\/\/www.imatag.com\/blog\/california-legislation-provenance-authenticity-and-watermarking-standards-act\" rel=\"nofollow noopener\" target=\"_blank\">New California regulations<\/a> requiring proof of content authenticity are an important step toward restoring trust but will not be sufficient. Users will need stronger media literacy to navigate an environment where seeing and hearing are no longer believing.<\/p>\n<p>\u2014 <a href=\"https:\/\/citris-uc.org\/people\/person\/camille-crittenden\/\" rel=\"nofollow noopener\" target=\"_blank\">Camille Crittenden<\/a>, executive director, <a href=\"https:\/\/citris-uc.org\/\" rel=\"nofollow noopener\" target=\"_blank\">CITRIS and the Banatao Institute<\/a><\/p>\n<p>\u2018Intelligence limits and the search for truth\u2019<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" width=\"448\" height=\"505\" data-ccwcag-attachment-id=\"137240\" data-ccwcag-attachment=\"{\" disable_page_edit=\"\" src=\"https:\/\/www.newsbeep.com\/us-ca\/wp-content\/uploads\/2026\/01\/gopnik_Cropped.jpeg\" alt=\"Alison Gopnik profile\" class=\"wp-image-137240\" style=\"width:270px;height:auto\"\/>Alison Gopnik<\/p>\n<p>I\u2019m expecting that,\u00a0in spite of the commercial pressures, we will realize that there is no such thing as general intelligence, artificial or natural. At the same time we may see progress toward more realistic models that engage and experiment with the external world, <a href=\"https:\/\/news.berkeley.edu\/2024\/08\/13\/why-studying-childrens-minds-could-help-us-build-better-ai\/\" rel=\"nofollow noopener\" target=\"_blank\">in the way that children do<\/a>.\u00a0<\/p>\n<p>My guess is that intrinsically motivated reinforcement learning systems \u2014 where the reward is finding the truth rather than getting a good score from humans \u2014 may make real progress in this regard.<\/p>\n<p>\u2014 <a href=\"https:\/\/vcresearch.berkeley.edu\/faculty\/alison-gopnik\" rel=\"nofollow noopener\" target=\"_blank\">Alison Gopnik<\/a>, professor of psychology<\/p>\n","protected":false},"excerpt":{"rendered":"In just a few years, AI has gone from a tech novelty to a society-altering force. It has&hellip;\n","protected":false},"author":2,"featured_media":132049,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[34],"tags":[2080,143,145,144,6080],"class_list":["post-132048","post","type-post","status-publish","format-standard","has-post-thumbnail","category-oakland","tag-artificial-intelligence","tag-oakland","tag-oakland-headlines","tag-oakland-news","tag-research"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/us-ca\/wp-json\/wp\/v2\/posts\/132048","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/us-ca\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/us-ca\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/us-ca\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/us-ca\/wp-json\/wp\/v2\/comments?post=132048"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/us-ca\/wp-json\/wp\/v2\/posts\/132048\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/us-ca\/wp-json\/wp\/v2\/media\/132049"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/us-ca\/wp-json\/wp\/v2\/media?parent=132048"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/us-ca\/wp-json\/wp\/v2\/categories?post=132048"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/us-ca\/wp-json\/wp\/v2\/tags?post=132048"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}