Usually, Data Points brings you the latest AI news, tools, models, and research in brief. But in today’s special New Year’s 2026 edition, you’ll find something different: a collection of our readers’ highest hopes for AI in 2026.
Want a sneak peek? We’ve got you covered:
- Fan Wu on automation and the smart home
- Ethan Flory on AI that knows when to stop
- Jen Gennai on the need for resilience
- Marsh Naicker on local multimodal models
- Navin on moving beyond the transformer
- Prabhu Rajagopalan on water conservation
- Asha Makur on AI for good
- Patrick Sieber on curing common cancers
- Egzon Baruti on improving memory
- Carlo Barth on developing person models
Fan Wu
My highest hopes for 2026 is that AI will be able to automate tasks like data extraction for work efficiency, and help to generate more insights for analysis on a small team scale; offer decent privacy protection to start using chatbot for entertainment and therapy use; and provide an initial smart home demo for every day tasks.
Ethan Flory
My highest hope for AI in 2026 is that, across products, platforms, models, and form factors, I start to notice automatic optimization for what is most important for me. What I mean is best illustrated by a quote from Edwin Chen of Surge AI, loosely paraphrased: AI might help you iterate 30 times to draft the perfect email over half an hour, and the model is eager to continue helping you improve it. But is that actually what you want? Or would you rather it tell you to stop after 5 minutes and send the email, that it's good enough and future improvements won't make a material difference. What I hope is that the objective function shifts towards actual human benefit rather than something closer to maximizing engagement.
Jen Gennai
My highest hope is that AI goes through its toughest test ever, yet proves resilient in preventing catastrophic harm. We know that in addition to AI's promise of helping humans address issues that they've never been equipped to tackle before – whether human longevity, scientific breakthroughs, or climate change mitigation – it is also not fully explainable, controllable, fair, or risk-free. As leading labs raise their own concerns about how bad actors could leverage their most advanced technologies for nefarious purposes, there is a non-zero risk of a catastrophic system failure or large-scale attack in 2026. However, a true testament to AI is not simply that it continues to gain momentum and integrate into more aspects of our everyday lives, nor that it avoids all major attack attempts, but that it shows its resilience in the face of even the worst attacks. In 2026, I hope that thanks to AI developers' responsible AI efforts that they can prove that advanced AI can be trustworthy and human-protecting by preventing or mitigating the effects of, serious attacks or manipulation – no matter how complex and insidious – from causing widespread human or societal harm.
Marsh Naicker
I hope that a powerful and trustworthy open-source multimodal AI model becomes available in 2026, ideally under 10 gigabytes in size and fast on Apple Silicon. This would allow people without an internet connection to benefit from AI. Above all, I hope AI progress is measured not just by technical metrics, but by how meaningfully it improves the lives of ordinary people, especially those who may never use a chatbot, for example by reducing or eliminating queues in hospitals.
Navin
I hope to see AI research being encouraged in a way that it tingles curiosity and creativity in people's minds so that they come up with new ideas instead of being stuck with the transformer architecture. This requires fundamental changes in academia for how research publication is rewarded and the choice of AI tools (Mojo language for example) that help build AI from scratch. Even corporations and governments need to create the luxury of time, to allow people to ponder creative solutions rather than be so driven by competition that it leaves barely any time for mental and physical rest.
Prabhu Rajagopalan
My highest hope for AI is that it should become less water intensive, and that the pathways to reducing water intensity should get clearer in 2026.
Asha Makur
The race for AI has created a lot of social problems – unemployment, fear of unemployment, brainwashing using deepfake scams, suicides etc. My heart breaks when the ChatGPT suicide story gets a lot less news cycle coverage and human attention than the data centers on Mars and other ambitious stories. Going forward, I would like to see the AI community use their collective competency and influence to make the world truly a better place. If videos need a warning "generated by AI, trust at your own risk." How are they different from cigarettes? What have we learned?
Egzon Baruti
My highest hope for AI in 2026 is efficient memory. Today, models remember everything or fewer things, but not the right thing!
Dr. Carlo Barth
In 2026, I'd like to see AI assistants move beyond "text in, text out" and treat who is speaking as first-class context. Human communication is grounded in an evolving model of the other person – their knowledge, goals, sensitivities, and likely interpretations – refined over repeated interactions. Most LLM training abstracts away stable identity and state, so systems can be fluent yet weak at sustained, personalized understanding. I hope we see architectures and training objectives that build compact, updatable "person models" alongside language modeling, enabling long-horizon alignment and usefulness without relying on ever-growing context windows.
Want to know more about what matters in AI right now?
Read last week’s special holiday edition of The Batch, which looks back at the most important AI stories of 2025.
Last week, Andrew Ng talked about the importance of structured learning through AI courses, hands-on practice in building AI systems, and occasionally reading research papers to enhance skills and inspire new ideas.
“I’ve heard some developers advise others to just plunge into building things without worrying about learning. This is bad advice! Unless you’re already surrounded by a community of experienced AI developers, plunging into building without understanding the foundations of AI means you’ll risk reinventing the wheel or — more likely — reinventing the wheel badly!”
Read Andrew’s letter here.
Other top AI stories of 2025 we covered in depth:
- Reasoning models, beginning with OpenAI’s o1 and DeepSeek’s R1, transformed the industry, solving bigger problems and setting new benchmarks.
- Meta’s hiring spree raised compensation for top AI engineers and executives, as big AI companies lure talent with huge pay.
- AI’s growth spurred infrastructure investment worldwide, leading to a significant data-center buildout.
- Software developers used more versatile AI-powered tools to write code, making agents write code faster and cheaper.
- China’s AI chip industry took root, as the country banned GPUs and TPUs from U.S. manufacturers to promote domestic chip development.
Looking Forward to 2026
That’s it for our special New Year’s edition of Data Points. We’ll be back with news on Monday, January 5th. Be sure to check out this week’s equally special New Year’s issue of The Batch, which rounds up the highest hopes of several AI luminaries for 2026. It also includes a special message from Andrew Ng. Also, be sure to check back in with us this week on December 31st, where we’ll have our second part of our special New Year’s edition of Data Points. And to all our readers we wish a very happy (and very hopeful) new year!
A special offer for our community
DeepLearning.AI recently launched the first-ever subscription plan for our entire course catalog! As a Pro Member, you’ll immediately enjoy access to:
- Over 150 AI courses and specializations from Andrew Ng and industry experts
- Labs and quizzes to test your knowledge
- Projects to share with employers
- Certificates to testify to your new skills
- A community to help you advance at the speed of AI
Enroll now to lock in a year of full access for $25 per month paid upfront, or opt for month-to-month payments at just $30 per month. Both payment options begin with a one week free trial. Explore Pro’s benefits and start building today!