AI Weekly: Claude 4.7, Token Costs, Open Models, Backlash, and Practical Ways to Use AI
In this weekend “Project Synapse” episode, the hosts review major AI developments, including Anthropic’s release of Claude Opus 4.7 (with rapid complaints about lying and token/cost issues) and discussion of the broader shift toward tighter token economics, enterprise budgeting pressure, and Microsoft’s evolving M365 licensing that bundles Copilot and agents. They note growing AI backlash ranging from local resistance to data centers and concerns about profitability to reports of attacks on Sam Altman’s home. The conversation covers open and Chinese models (agentic coding and multimodal image generation) and the strategic impact of open weights. They also highlight real-world uses: automating documentation, internal Q&A knowledge bases, customer service (including Starlink using Grok), research and editing workflows, book marketing, document drafting, email/search, and accounting/expense automation—while emphasizing hallucinations and verification.
Hashtag Trending would like to thank Meter for their support in bringing you this podcast. Meter delivers a complete networking stack, wired, wireless and cellular in one integrated solution that’s built for performance and scale. You can find them at Meter.com/htt
00:00 Sponsor and Welcome
01:18 Lightning Round Setup
02:34 Claude Opus 4.7 Reactions
08:22 Token Costs and Enterprise Pricing
12:45 Valuations Profit and UBI Debate
19:14 Backlash Data Centers and Supply Chain
31:29 Copilot Hallucination at Work
33:32 China Agentic Coding Model
37:43 Practical AI Wins and New Threats
44:18 AI Crime Scale
45:24 Everyday AI Wins
48:29 Lightning Round Demo
50:17 Open Source Shockwave
52:32 Deepfake Son Dilemma
58:20 AI Marketing Playbook
01:02:23 Editing and Fact Checks
01:05:09 Ethics and Authenticity
01:16:27 AI First Checklist
01:22:28 Tooling Gets Better
01:25:35 Wrap and Sponsor
