DeepLearning.AI "The Batch" - New Year Special Issues
Introduction
Purpose
This analysis examines seven years of New Year prediction articles from DeepLearning.AI's newsletter "The Batch" (2020–2026) to understand what AI leaders believe is important—and, more revealingly, what they overlook.
By extracting and categorizing 44 predictions from prominent researchers and executives, we can identify which human needs the AI community is actively working to address, and which fundamental needs remain outside the field's collective attention.
Method
- Data Collection: Extracted subsection titles and contributor names from seven annual "hopes and predictions" issues of The Batch newsletter.
- Prediction Extraction: For each subsection, identified the author's core prediction, rephrasing where necessary to remove jargon while preserving intent. Each prediction was condensed to 10 words or fewer.
- Human Needs Mapping: Categorized predictions against a framework of 15 fundamental human needs. Five needs emerged as addressed by the articles (Trust, Belonging, Agency, Understanding, Progress); ten remained unaddressed.
- Gap Analysis: For each unaddressed need, brainstormed potential AI predictions, evaluated each for novelty and feasibility, and selected the most promising prediction to illustrate how AI could address that need in coming years.
Key Finding
AI leaders focus heavily on capability and ethics but largely ignore existential human needs—purpose, rest, authenticity, legacy, and transcendence. This suggests opportunity for differentiated AI products that address the deeper questions humans ask.
Predictions by Year
| Year | Issue # | Seq | Theme | Contributor | Subsection Title | Prediction Summary |
|---|---|---|---|---|---|---|
| 2020 | 20 | 1 | High Hopes for 2020 | Anima Anandkumar | The Power of Simulation | "We will see more research in simulation" |
| 2020 | 20 | 2 | High Hopes for 2020 | Oren Etzioni | Tools For Equality | "AI will tangibly benefit disadvantaged populations" |
| 2020 | 20 | 3 | High Hopes for 2020 | Chelsea Finn | Robots That Generalize | "We will see progress on models that generalize" |
| 2020 | 20 | 4 | High Hopes for 2020 | Yann LeCun | Learning From Observation | "Self-supervised methods will learn from video" |
| 2020 | 20 | 5 | High Hopes for 2020 | Kai-Fu Lee | AI Everywhere | "AI will penetrate our personal and business lives" |
| 2020 | 20 | 6 | High Hopes for 2020 | David Patterson | Faster Training and Inference | "Novel AI hardware will bear fruit" |
| 2020 | 20 | 7 | High Hopes for 2020 | Richard Socher | Boiling the Information Ocean | "Summarization will take important steps forward" |
| 2020 | 20 | 8 | High Hopes for 2020 | Dawn Song | Taking Responsibility for Data | "2020 will be the year we build responsible data economy" |
| 2020 | 20 | 9 | High Hopes for 2020 | Zhi-Hua Zhou | Fresh Methods, Clear Guidelines | "AI will bring positive changes to everyday lives" |
| 2021 | 72 | 1 | Onward to 2021 | Ayanna Howard | Training in Ethical AI | "AI community will reach consensus on ethical AI" |
| 2021 | 72 | 2 | Onward to 2021 | Fei-Fei Li | Invigorating the U.S. AI Ecosystem | "U.S. government will commit to support AI" |
| 2021 | 72 | 3 | Onward to 2021 | Matthew Mattina | Life-Saving Models in Your Pocket | "Neural nets on small devices will improve lives" |
| 2021 | 72 | 4 | Onward to 2021 | Harry Shum | Assisted Artistry | "AI will create tools to unleash creativity" |
| 2021 | 72 | 5 | Onward to 2021 | Ilya Sutskever | Fusion of Language and Vision | "Language models will become aware of visuals" |
| 2022 | 125 | 1 | Ring in the New | Abeba Birhane | Clean Up Web Datasets | "Work cleaning datasets becomes mainstream AI" |
| 2022 | 125 | 2 | Ring in the New | Wolfram Burgard | Train Robots in the Real World | "Roboticists will train robots in real-world" |
| 2022 | 125 | 3 | Ring in the New | Alexei Efros | Learning From the Ground Up | "AI learning from raw data will progress" |
| 2022 | 125 | 4 | Ring in the New | Chip Huyen | AI That Adapts to Changing Conditions | "More companies will use real-time ML inference" |
| 2022 | 125 | 5 | Ring in the New | Yoav Shoham | Language Models That Reason | "NLP re-embraces symbolic reasoning" |
| 2022 | 125 | 6 | Ring in the New | Yale Song | Foundation Models for Vision | "Foundation models will pretrain on unlabeled video" |
| 2022 | 125 | 7 | Ring in the New | Matt Zeiler | Advance AI for Good | "Public gains appreciation of AI benefits" |
| 2023 | 177 | 1 | Get Ready for 2023! | Yoshua Bengio | Models That Reason | "We will see progress toward human-like reasoning" |
| 2023 | 177 | 2 | Get Ready for 2023! | Alon Halevy | Your Personal Data Timeline | "AI that unifies all your personal data will help you" |
| 2023 | 177 | 3 | Get Ready for 2023! | Douwe Kiela | Less Hype, More Caution | "We will see work on alignment and attribution" |
| 2023 | 177 | 4 | Get Ready for 2023! | Been Kim | A Scientific Approach to Interpretability | "We will enable richer human-machine collaborations" |
| 2023 | 177 | 5 | Get Ready for 2023! | Reza Zadeh | Active Learning Takes Off | "Active learning will take off" |
| 2024 | 229 | 1 | Generating 2024 | Anastasis Germanidis | New Tools to Tell New Stories | "Video generation quality improves rapidly" |
| 2024 | 229 | 2 | Generating 2024 | Sara Hooker | Prioritize Inclusion | "AI will improve for under-resourced languages" |
| 2024 | 229 | 3 | Generating 2024 | Percy Liang | Transparency for Foundation Models | "We will start to reverse transparency decline" |
| 2024 | 229 | 4 | Generating 2024 | Sasha Luccioni | Respect for Human Creativity and Agency | "New mechanisms will give creators more control" |
| 2024 | 229 | 5 | Generating 2024 | Pelonomi Moiloa | Smaller Models That Learn More From Less Data | "We will build smaller, smarter models" |
| 2024 | 229 | 6 | Generating 2024 | Kevin Scott | Be Prepared for Another Year of Exponential Growth | "Many new AI apps and tools will emerge" |
| 2025 | 282 | 1 | 2025 Beckons | Hanno Basse | Generative AI for Artists | "Generative AI will help people be creative" |
| 2025 | 282 | 2 | 2025 Beckons | David Ding | Generated Video With Music, Sound Effects, and Dialogue | "Models will produce video with full audio" |
| 2025 | 282 | 3 | 2025 Beckons | Joseph Gonzalez | General Intelligence | "Focus will turn to showing real value" |
| 2025 | 282 | 4 | 2025 Beckons | Albert Gu | More Learning, Less Data | "Models will learn more from less data" |
| 2025 | 282 | 5 | 2025 Beckons | Mustafa Suleyman | Agents of Action | "AI will act autonomously on our behalf" |
| 2025 | 282 | 6 | 2025 Beckons | Audrey Tang | AI That Unites Us | "AI will power platforms that promote empathy" |
| 2026 | Special | 1 | Agents of 2026 | David Cox | Open Source Wins | "Open-source AI will flourish and win" |
| 2026 | Special | 2 | Agents of 2026 | Adji Bousso Dieng | AI for Scientific Discovery | "AI transitions to catalyst for discovery" |
| 2026 | Special | 3 | Agents of 2026 | Juan M. Lavista Ferres | Education That Works With — Not Against — AI | "Education will develop reliable AI-era evaluation" |
| 2026 | Special | 4 | Agents of 2026 | Tanmay Gupta | From Prediction to Action | "AI agents will handle vague, open-ended goals" |
| 2026 | Special | 5 | Agents of 2026 | Pengtao Xie | Multimodal Models for Biomedicine | "Community will build multimodal biomedical models" |
| 2026 | Special | 6 | Agents of 2026 | Sharon Zhou | Chatbots That Build Community | "AI will bring people together" |
Human Needs Addressed
| Human Need | What People Say | Addressed By |
|---|---|---|
| Trust | "Can I believe what this system tells me?" | 2020.8, 2021.1, 2022.1, 2023.3, 2024.3, 2024.4 |
| Belonging | "I feel left out of the benefits" | 2020.2, 2024.2, 2025.6, 2026.6 |
| Agency | "I don't have time for everything" | 2021.4, 2025.1, 2025.5, 2026.4 |
| Understanding | "I want to be truly understood" | 2020.4, 2021.5, 2022.5, 2023.1, 2023.2, 2023.4 |
| Progress | "We're stuck on the hard problems" | 2020.6, 2021.3, 2024.5, 2025.4, 2026.2, 2026.5 |
Human Needs Not Addressed
| Human Need | What People Say | Prediction |
|---|---|---|
| Purpose | "What am I here for?" | AI will help discover purpose by analyzing what energizes you |
| Privacy | "I want to be left alone" | AI agents will negotiate privacy terms on your behalf |
| Security | "Will I be okay financially?" | AI will provide early warning of personal financial risks |
| Rest | "I'm exhausted" | AI will protect your time by filtering low-value demands |
| Authenticity | "I want to be my real self" | AI will help you find communities that embrace your real self |
| Legacy | "Will I matter after I'm gone?" | AI will preserve your stories and wisdom for future generations |
| Simplicity | "Life is too complicated" | AI will absorb complexity so you only face simple choices |
| Nature | "I need to get outside" | AI will nudge you outdoors by sensing when you need nature |
| Justice | "This isn't fair" | AI will surface hidden inequities ordinary people can't see |
| Transcendence | "There must be more than this" | AI will facilitate peak experiences through personalized art |
Source URLs
- Issue 20 (2020)
- Issue 72 (2021)
- Issue 125 (2022)
- Issue 177 (2023)
- Issue 229 (2024)
- Issue 282 (2025)
- 2026 Special
Analysis by Claude Opus 4.5 (Thinking) via Windsurf. January 2026.
posted by Michael Osofsky at
2:03 AM
0 Comments:
Post a Comment
Subscribe to Post Comments [Atom]
<< Home