Browse Comments — Clean (de-noised)

Microsoft restricted claude usage , however they haven't shifted to human(junior devs) they have started to use their own co-pilot

Token size matters and context matters while using AI these companies trained the AI Also no one talks about completion productivity in scales Always they create hype in public One way of updating their presence and competitions Cost of products increased in the name of AI enabled.... What to say

If your work has been negatively impacted by AI, we want to hear from you: aicommonsproject.org or follow our page for more information

Anil Agarwal Agreed

SAURABH SINGH Agree. Recently, I asked someone that how are they using AI with unlimited tokens at your plate. His response was very simple :) - “I am just using it for all without any tokens consumption analysis “ I asked him why so- he says because we have unlimited tokens :) why track? The above mindset is actually making people lazy enough to think and be a better Developer+researcher. I saw the way suddenly people have started using AI tools without any knowledge, understanding of what the documentation says on how to use it effectively is causing “AI Fatigue” . People are getting tired because of too much prompting, session management etc. I believe if companies start asking the details about the bill and usage proportional to there productivity , most of people will get stuck on how to present and explain it as they never thought of. What to do? Make people around you in your office aware of things on how to use such tools effectively. Have sessions about it. Have a role called AI champion, who experiments and broadcasts his findings org wide. Have one meeting internal to the team where user explains his usage and metrics and how is he optimising it. Not sure everyone will follow but if few catch up it still a win.

Big companies are playing the race of reducing humans which is unlimitedly have reverse impact, We need to set some governance for usages then we can make effective utilization.

This is the inevitable collision between brute-force API calling and enterprise reality. The mistake wasn't adopting AI, it was treating LLMs like an infinite resource rather than an engineering constraint. Throwing raw, un-optimized context at expensive foundational models is the modern equivalent of leaving cloud instances running at peak capacity 24/7. True AI strategy isn't just about what the model can do, it's about building intelligent orchestration, semantic caching, and specific routing to ensure the unit economics actually scale.

SAURABH SINGH And that they already knew, let's be honest.

Cory Gibson That, or widespread belt-tightening by the growing masses of "replaced" workers will squeeze the economy because of how much money is no longer being spent on the goods and services these companies provide.

Whatever. It has electrolytes!!!

Your point about token based pricing turning AI into a costly tool highlights why governance around usage is becoming as important as the technology itself. How are teams balancing the speed gains with strict cost monitoring practices?

Well AI had a ton of certificates and skill sets gained over the years, went through multiple appraisel cycles. It is expected.Convinience and Ignorance are a bliss - until they hits.

I am convinced the prices will only increase as the constraints surrounding AI services tighten (energy, water, real-estate, chips, etc..) but while the price was always going to be high once the subsidies started to whilttle away what i think is interesting is how ROI of AI will be measured going forward. For example, human hourly rate vs AI task completion is not apples to apples and may not necessarily spell the replacement of a persons job but if that task would have taken a team of Data Scientists a week then perhaps a $2k price seems fair for a quick run... the question might become, what work should the team delegate this week? This of course requires organisational understanding on AI friendly work, forward planning and a strategic adoption where it makes sense.

Getting more expensive and not up to the same human level - not at all. I hope now everybody will realize that Humans will do better with the help of AI, instead of AI replacing humans. Humans first

Corporations want ai to work so they can lay-off the whole team, eliminating any class contributing to the economy, this is pandora's box

it’s a multiplier, not a replacement. And, if you don’t monitor or supervise the purpose and usage, you will always get a suprise end of every month.

Interesting perspective SAURABH SINGH. This is a good reminder that technology adoption and economic viability are not always the same thing. AI can dramatically increase leverage in the right workflows, but scaling usage without clear ROI, governance, and operational discipline can quickly turn efficiency gains into cost inflation. The long-term winners will likely be the companies that treat AI as a strategic multiplier for human capability — not simply as a replacement strategy.

Shocking.....companies didnt do research on their budgeting, their spend, their replacement costs, their finances at all before spending. Sounds like their boards need fired for approving.

The problem isn’t that AI is expensive. The problem is most companies are layering AI on top of broken workflows and expecting magic. Giving every employee Claude access is not an AI strategy. That’s just enterprise autocomplete with a bigger cloud bill. The real shift happens when finance and operations become AI-native from the workflow layer itself: reconciliations that continuously learn, approvals that understand operational context, forecasting tied to live execution, and systems that compound intelligence over time. That’s the direction we’re thinking about deeply at FinoLusion AI. Otherwise companies are just burning tokens to accelerate the same inefficiencies they already had.

Because we still in exploration stage, and people still learning on how to use it. Eventually token consumption will drop and cost stabilize