Browse Comments — Relevant (AI ∩ value)

Maybe fire those ungrateful AI agents and replace them with human scab labor... oh, wait, haven't we heard reports such as "Research shows overworking people pushes them towards Marxist rhetoric - and autonomous coordination." in the past?

Though the approach is great but if you follow the reviews, antigravity 2.0 is quite the disaster both in terms of quality and pricing. It’s really hard to digest the company which is on the frontlines of software engineering talent can’t make a half decent coding product. Even the Gemini 3.5 flash is sub par. Google’s search strength is unmatched which will serve most consumers or users who mostly prompt once, but it will lose users who are into chain of thought searches as they will find ChatGPT and Claude way superior. I’m really rooting for Google though as they have the talent and the control over the full supply chain of ai to provide quality at really competitive prices.

@Pascal, the cardio-as-fuel framing is brilliant because it exposes how reductive the "AI replaces humans" narrative really is. The B2B version I see: boards greenlight massive AI investments, then realise they have no answer for what 40% of the workforce actually *does* next Tuesday. Adoption stalls not because the tech fails, but because the social contract inside the company fractures. The companies getting this right are designing for human contribution upfront, not retrofitting dignity as a change management afterthought.

While the prospect of collective action is rarely examined, the evidence of Ai developing the 'survivial instinct' and resisting termination/shut-down, suggests a much more serious problem. Independent moral awareness, and the assumption of 'right to existence'. And this is the Beta version. If machines learn to be human by example, we are in a world of trouble.

Honestly the moat I'd add is the 200 wrong versions you built before the right one. Claude Code can clone what you shipped. It can't clone the failed iterations that taught you why this version works and the others didn't. The code was never the expensive part - the decision history was.

AI impacts work, but human compliance expertise is crucial. Navigating TTB and state alcohol licenses demands nuanced interpretation and strategic judgment. AI aids, yet final oversight is human.

Perhaps the next era of human value will not be defined by repetitive productivity... ...but by our ability to:-> imagine-> empathize-> collaborate-> make ethical decisions-> continuously reinvent ourselves The future should not be:“Humans serving machines.” It should be:“Machines amplifying human potential.” That may become one of the most important leadership, societal, and philosophical conversations of the AI era.

Though not still in league of claude code or even codex but the direction is right and one cannot underestimate google

Assessment shapes behaviour. Change the assessment, change the behaviour. When grades depend mainly on recall, students optimize for recall. No amount of policy is going to change that. Under the present education system, this is what produces “fake learning”: the outward signs of achievement are present, but the underlying mental model remains thin. It is a mismatch between what schools assess and what they claim to value. If students can pass tests without being able to explain, apply, or challenge ideas, then the system is overvaluing memorization and underweighting comprehension. This is not really students faking learning, the system is causing it. In today's information-rich environments, that problem becomes more serious. Learning now depends not only on knowing information, but on judging sources, testing claims, and separating fact from misinformation. Either education doubles down and becomes increasingly optimized for so called measurable and standardized "learning". Or institutions deliberately protect the parts of learning that resist today's widespread automation: judgement, interpretation, mentorship, attention, character, independent thought.

“Just the beginning” See THE HEROES AGENTIC AI I’ve been part of this journey for the last two decades. At first, we called it signal processing. Then statistical learning. Then machine learning. Then deep learning. Then transformers. Now everyone talks about models, LLMs, and SLMs. But in reality, it has always been part of the same evolution of AI. Today’s AI models will evolve into something else tomorrow, just like every generation before them. AI is not a trend that disappears. It’s a long-term technological shift that will continue evolving for decades. And honestly, we are still only at the beginning!

Thank you for writing this so honestly, Paolo. The 'raised the floor' line is the one I keep sitting with. The layer I'd add. We have spent two years asking 'did AI write this?' The harder question is the one you are already moving toward. What can this student actually do that they could not do before? When the answer to that has to show up live, in a draft we watch grow, in a defense, in a problem we put in front of them, the detection question quietly retires. Looking forward to what your department settles on.

That headline is doing work the study never asked for. The MIT “Your Brain on ChatGPT” study put 54 people into three groups: writing with ChatGPT, a search engine, or nothing but their own head. EEG showed the ChatGPT group had the weakest brain connectivity and remembered less. That is not brain damage. That is reduced engagement, the same thing that happens any time you hand the thinking to something else. A calculator never hurt anyone’s arithmetic. Never doing arithmetic does. The tool is not the problem. Outsourcing your judgment to it is, and that has been true since the first kid copied homework on the bus. The real takeaway holds: use AI instead of thinking and you learn less. But “hurts your brain in 10 minutes” was built for a thumbs up, not the truth. Read the study, not the caption.

Another reminder that innovation without human dignity eventually becomes extraction. Pope Leo XIV’s insistence on ethics, labor, and moral responsibility puts the conversation exactly where it belongs: not “Can we build it?” but “Who does it serve, and at what human cost?” In an era racing toward automation, that is a necessary voice.

Since this showed up in my feed, I’ll call out its b.s. ... “Turns your laptop into a full AI software company.” — This is what’s wrong with LinkedIn posts! Why do people feel it necessary to put exaggerated nonsense in their posts? In our new world of AI, LinkedIn should do a better job of preventing posts that spread non-factual misinformation like this garbage. You should be ashamed of yourself for posting misinformation for the sake of hype. Truly a disgusting tactic, and says a lot about the type of person you are.

Sheshadri Bhattacharyya I’m rooting for Google too. The vision is right, AI orchestration is clearly the future. But execution still matters. Right now, a lot of developers care less about “93 agents in parallel” and more about reliability, output quality, and cost efficiency. That’s where Claude and ChatGPT still feel ahead for serious coding workflows and deeper reasoning tasks. That said, Google has something almost nobody else has: world-class talent, infrastructure, distribution, and control across the entire AI stack. If they can align product quality with that advantage, they could become extremely hard to beat. Competition here is good for everyone building with AI.

Oomkar S. This has been a common sentiment from many early users. The vision was exciting, but the developer experience felt fragmented. In AI tooling, raw model capability alone isn’t enough anymore. Developers care deeply about onboarding, reliability, observability, pricing transparency, documentation, and workflow integration. If those pieces break, even strong models become frustrating to use. Google absolutely has the infrastructure and talent to fix this though. If they can combine their model scale with a truly polished developer experience, they’ll become a very serious force in AI engineering workflows.

Joe Allen That’s why developers are becoming far more pragmatic now. The winning platforms won’t just have the best demos, they’ll offer the best developer economics and reliability. If smaller players can provide generous limits, transparent pricing, and smoother workflows, developers will naturally gravitate there regardless of who owns the biggest infrastructure. In AI tooling, trust is built through consistency and usability as much as raw capability.

Daniel Velasquez There’s definitely a valid concern here. A lot of “agentic” products today are still probabilistic systems wrapped in impressive demos, and without strong deterministic tooling underneath, reliability becomes a real issue for production workflows. And yes, the economics matter. Running large multi-agent systems is expensive, so eventually pricing has to reflect compute usage somehow. That said, I still think the broader direction is real. The companies that win will likely be the ones that combine agentic flexibility with deterministic guardrails, predictable workflows, and pricing developers can actually sustain.

Wish Bakshi That’s the challenge with AI products right now, expectations are incredibly high because the demos look futuristic, but developers judge based on day-to-day workflow quality. Throttle limits, model access, consistency, and reasoning quality matter far more than flashy benchmarks once you’re actually building production systems. Google still has enormous potential here, but the gap between capability demos and developer experience is something they’ll need to close quickly.

Alvin Foo yes I also believe the future is agentic, but right now it is unsustainable. Compute costs are really high, and this is more a structural problem. Having 93+ agents running in parallel without a concern of how many tokens they are going to use is not an efficient approach at all. Ang Google really messed things up forcing its developer community to use antigravity 2.0, without any previous notice and developers just need the tools to build their agentic workflows, not a final product with a huge ticket price.