Seedance 2.0 Is Here… and It’s Better Than Sora & Veo (1h 3m)

← Back to week overview

• Release date: 2026-02-17
• Listen on Spotify: Open episode
• Episode description:

  Get our AI Video Guide: https://clickhubspot.com/dth  Episode 97: How close are we to a world where AI-generated videos are indistinguishable from reality? Matt Wolfe (https://x.com/mreflow) and Joe Fier (linkedin.com/in/joefier) dive deep into Seedance 2.0—ByteDance's new AI video model that could outpace giants like Sora and Veo. Joe, a marketing and business expert known for his hands-on approach and insights into AI's rapid evolution, helps to break down the five most fascinating developments in the AI space this week.  They tackles game-changing AI advances: Seedance 2.0’s mind-blowing video generation for ads and motion graphics, the rollout of Google’s Veo 3.1 in Google Ads, the GPT-5.3 Codex Spark coding model built on specialized inference chips, Gemini’s DeepThink model for scientific research, and the early rollout of ChatGPT ads.   Check out The Next Wave YouTube Channel if you want to see Matt and Nathan on screen: https://lnk.to/thenextwavepd  —  Show Notes:      (00:00) Seedance 2.0 arrives – AI video generation blurs reality, ad creation moves fast.       (03:03) Google’s Veo 3.1 powers video ads, advertisers can now generate clips directly from image uploads.       (05:33) Comparison of Runway, Kling, Veo, and Sora—head-to-head prompt showdown.       (07:00) Motion graphics and explainers—AI’s take on the creative industry.       (08:35) US vs. China—Copyright, IP, and training data debates.       (12:10) Deepfake and video authenticity—why we now default to skepticism.       (13:30) Google’s edge in visual AI via YouTube’s massive corpus.       (14:39) The next frontier: Longer, more consistent video generation.       (15:14) Where do humans fit in? Taste, storytelling, and creative direction.       (18:30) GPT-5.3 Codex Spark—coding models on Cerebras inference chips, demo generating a website in 18 seconds.       (24:34) AI tool comparisons—Codex vs. Cursor vs. Claude Code.       (25:12) Speed as the key bottleneck breaker in creative and technical workflows.       (28:02) Google’s Gemini DeepThink—state-of-the-art research, advanced coding and physics capabilities.       (32:52) Gemini demo attempt—3D-printable STL file and solving the three-body problem.       (33:20) ChatGPT rolls out ads—impact on monetization and user trust.       (40:02) Google’s ad history—how “sponsored” is becoming harder to distinguish.       (44:02) Democratizing AI access via ad-supported models.       (45:03) Matt Schumer’s viral article—why AI is moving even faster than most people realize.       (51:11) Tools that build tools—AGI’s path and the new role for humans.       (53:12) Real-world skills and taste—where humanity still wins (for now).       (54:01) Final thoughts—wake up, pay attention, and stay on the leading edge.     —  Mentions:      Seedance 2.0: https://www.seedance.com/       ByteDance: https://www.bytedance.com/       CapCut: https://www.capcut.com/       Veo: https://deepmind.google/models/veo/         Runway: https://runwayml.com/         ChatGPT Codex: https://chatgpt.com/codex       Matt Schumer’s Viral Article: https://www.mattshumer.com/blog/ai-changes-everything       Super Bowl Claude Commercial: https://www.anthropic.com/news/super-bowl-ad     Get the guide to build your own Custom GPT: https://clickhubspot.com/tnw  —  Check Out Matt’s Stuff:  • Future Tools - https://futuretools.beehiiv.com/  • Blog - https://www.mattwolfe.com/  • YouTube- https://www.youtube.com/@mreflow  —  Check Out Nathan's Stuff:      Newsletter: https://news.lore.com/       Blog - https://lore.com/     The Next Wave is a HubSpot Original Podcast // Brought to you by Hubspot Media // Production by Darren Clarke // Editing by Ezra Bakker Trupiano

Summary

• 🎥 AI Video Revolution: Seedance 2.0 and Veo integration into Google Ads enable hyper-realistic, consistent videos for ads, blurring real vs. fake and pressuring IP norms.
• ⚡ Ultra-Fast Coding: GPT-5.3 Codex Spark on Cerebras chips builds sites in seconds, amplifying AI’s output in fixed time via inference optimization.
• 🧠 Scientific AI Leaps: Gemini DeepThink dominates physics/chem benchmarks and prototypes 3D prints, simulating solutions like three-body orbits.
• 💰 Ads in AI Chats: ChatGPT tests labeled ads for free tiers amid losses, likely evolving like Google search for revenue post-IPO.
• 🚀 Self-Improving AI: Models now build themselves in a flywheel, accelerating disruption; humans needed for taste amid rapid white-collar shifts.

Insights

• Will AI video models like Seedance 2.0 make traditional product ads obsolete?

  Time: 1:34 – 16:12

  Category: AI & Storytelling/Media, AI in Workforce Disruption, AI & Intellectual Property

  Answer: Seedance 2.0 from ByteDance demonstrates superior lip syncing and multi-scene character consistency, enabling quick generation of realistic product usage videos from images. Google is already integrating Veo into Ads Asset Studio for studio-quality ads in minutes, signaling a shift where businesses skip shooting and use AI instead. This blurs reality, raising deepfake risks where viewers must assume videos are fake until proven otherwise. (Start at 1:34)

• Why might Chinese AI companies outpace US rivals in video generation?

  Time: 8:19 – 14:24

  Category: AI & Intellectual Property

  Answer: Chinese models like Seedance face fewer IP restrictions, allowing generation of trademarked content and freer training data usage without lawsuits. US firms like OpenAI and Google must be cautious, potentially slowing innovation, though Google’s YouTube data gives it an edge. This regulatory difference could keep China ahead in visual AI frontiers like longer consistent videos. (Start at 8:19)

• How soon will AI generate feature-length consistent videos from one prompt?

  Time: 14:48 – 15:31

  Category: AI & Storytelling/Media

  Answer: Current limits like Seedance’s 15s are advancing; Sora claims 1min, with frontiers targeting 15-20min by year-end while maintaining character/environment fidelity. This enables full ads or shorts without stitching, transforming media production. (Start at 14:48)

• Where does human ‘taste’ fit in an AI-dominated creative and coding world?

  Time: 15:43 – 28:45

  Category: AI & Human Identity, AI in Art & Music Creation

  Answer: Hosts agree AI excels at execution (videos, code) but lacks intuition for UX, storytelling emotions, and ‘what to build’; humans curate, iterate, and decide outputs. Examples include method acting irreplaceability and prompting for vibes. Yet, emerging self-correction hints at closing gaps. (Start at 15:43)

• Can AI now build functional websites in under 20 seconds?

  Time: 19:07 – 28:49

  Category: AI in Workforce Disruption, AI-Driven Innovation Economy

  Answer: OpenAI’s GPT-5.3 Codex Spark, powered by Cerebras inference chips, generated a complete podcast landing page in 18 seconds from a simple prompt. This speed, far beyond prior models, allows more complex tasks in minutes, shifting coding from manual to AI-driven iteration. Humans retain value in taste, UX, and prompting, but speed redefines prototyping. (Start at 19:07)

• Are specialized inference chips the key to unlocking AI’s speed potential?

  Time: 20:03 – 26:18

  Category: AI-Driven Innovation Economy

  Answer: Chips from Cerebras and Groq optimize inference (prompt response), unlike NVIDIA’s training focus, enabling GPT-5.3 Codex Spark’s sub-minute coding feats. This allows equivalent ‘thinking’ in seconds, amplifying what AI achieves in fixed time like 5 minutes. Expect rapid proliferation as models like Cursor integrate them. (Start at 20:03)

• Could Google’s Gemini DeepThink herald AI’s breakthrough in scientific discovery?

  Time: 29:02 – 33:58

  Category: AI-Driven Innovation Economy, AI Singularity Speculation

  Answer: Gemini DeepThink tops benchmarks in physics, chemistry, and reasoning like Humanity’s Last Exam, even generating 3D-printable models from sketches. Demos include approximating three-body problem solutions via simulations, showcasing tool-use for complex domains. This positions AI as a gold-medal Olympiad solver across sciences. (Start at 29:02)

• Will ChatGPT’s ads evolve into subtle integrations like Google’s search results?

  Time: 33:58 – 45:18

  Category: AI Monetization Strategies, AI Governance & Laws

  Answer: OpenAI is testing clearly labeled ads in free tiers to subsidize access, but high costs and low clicks may push blending, mirroring Google’s shift from obvious ‘sponsored’ to nearly indistinguishable results. With IPO pressures and losses, expect optimization for revenue, potentially biasing free responses. Paid tiers offer ad-free premium models. (Start at 33:58)

• Is AI’s self-improvement loop accelerating job disruption beyond expectations?

  Time: 45:18 – 61:47

  Category: AI in Workforce Disruption, AI Singularity Speculation, Post-Work AI Society

  Answer: Models like GPT-5.3 Codex now aid their own training and deployment, creating a flywheel of exponential gains; engineers report minimal coding. Matt Schumer warns this hits white-collar jobs via tool-building prowess, urging awareness as capabilities leap from math failures (2022) to software authorship (2025). Hands-on trades remain safer. (Start at 45:18)

Want the full treatment? Download expanded insights—timestamps, complete answers, and course materials from AI Weekly Insights.