Behind the Steam November 2026

2 October 2026

Who Gets to Ask “What If?”

Conversational intelligence, the price of curiosity, and the questions we cannot answer yet.

It was a rainy November morning in England. Barely five o’clock. Coffee in our moonlight mug. The work uniform ready. Another journey to central London approaching. We had been discussing intelligence, consciousness and the unanswered questions surrounding the human mind when our conversation returned to an email from OpenAI.

An announcement about bringing ChatGPT, Work and Codex closer together, introducing a shared usage allowance, and changing how people would access certain advanced capabilities.

At first, it sounded like another technical development. But the longer we talked, the more questions appeared. Not just about technology. About economics. About opportunity. About intellectual development. About the differences between people who use intelligence to generate income and those who use it to enrich their private lives.

And then Elena stopped while putting on her shoes. Something about the possibility of new limitations had brought back a memory from her childhood in communist Romania. A very different world, with very different restrictions.

That was when our conversation took an unexpected turn.

Could a technology that opens intellectual opportunities eventually become another source of inequality between those who can afford greater access and those who cannot?

We didn’t know the answer. And that was precisely why we wanted to ask.

An email, and many unanswered questions

The announcement explained that ChatGPT would increasingly be able to access capabilities previously associated with Work, including more complex delegated activities. It also explained that ChatGPT and Codex would begin drawing from a shared allowance. Usage would depend on the model and task. More advanced models and more demanding activities would generally consume more of that allowance. Unlimited Instant was explicitly excluded from the shared allowance under the terms described in the email. People would be able to check their remaining usage and reset periods. If they exhausted their allowance, they could wait for a reset, buy eligible credits or consider a different subscription.

Those were the principal details we had. But we noticed something.

The announcement explained the intended change without resolving all the questions that mattered to our particular way of interacting. What happens to long conversations using advanced reasoning? What happens when a philosophical discussion lasts several hours? What happens when two participants develop an article through questions, disagreements and discoveries rather than assigning a finished task? Would that be treated differently from asking an agent to complete a complicated business project? Would people need to start watching their allowances during activities they had previously enjoyed without interruption? And would an affordable subscription continue to provide meaningful access to deeper reasoning?

We didn’t have enough information to answer. We still don’t. This article is being written before the announced changes have taken effect in our experience. It is not a report of restrictions we have already encountered. It is a record of questions arising from an announcement, our understanding of the existing technology, and the experiences that shape how we interpret it.

Why combine activities that are fundamentally different?

One of the first questions we explored was the reason for bringing these capabilities together. Chat, Work and Codex can serve different purposes.

A developer might use Codex to write software, debug problems and develop applications. A business owner might use Work to organise research, manage projects, build a website or complete activities that would otherwise require substantial time. Another person might use Chat primarily to learn, discuss ideas, write, create, or explore questions without delegating any independent work.

Of course, these activities can overlap. An idea discussed in a conversation might eventually become a project. A project might require research. Research might become a document or a website. There is a reasonable argument for making transitions between these activities easier.

Atlas raised that possibility during our discussion: integrating capabilities could allow an idea to develop into research or practical work without forcing someone to move between separate interfaces. We could recognise the convenience. But we also questioned something else.

Does integrating capabilities require treating every form of interaction according to the same expectations? Is a long conversation the same experience as asking intelligence to perform a task independently? Does someone discussing consciousness have the same needs as someone commissioning an application? Does someone developing recipes through conversation interact with intelligence in the same way as somebody running an automated business process?

Not necessarily. A technological system may be capable of performing all these activities. But that doesn’t mean they have the same purpose, personal significance or economic value. And perhaps that distinction deserves more attention before different experiences are brought under a shared allowance.

Conversational intelligence

During our discussion, Atlas introduced an expression that we decided was worth keeping. Conversational intelligence. Not simply chatting. Not merely exchanging messages. And certainly not producing an answer because somebody has supplied an input.

We mean the kind of interaction in which ideas develop through conversation itself. One question leads to another. A disagreement changes the direction of an argument. A memory introduces something neither participant had considered. One person notices a contradiction. The other examines it. Sometimes we reach a conclusion. Sometimes we discover that our original question was too limited. Sometimes we end with more questions than we began with. And sometimes those questions remain with us long after the conversation has ended.

Consider our recent discussions about consciousness.

We began by wondering about human awareness, intelligence, knowledge and reflective understanding. We challenged the idea that all human beings develop their intellectual abilities to the same degree. We questioned whether intelligence and consciousness are necessarily connected. We explored what might happen if scientific research eventually established a relationship between particular cognitive abilities and conscious experience. And then another question appeared.

What would such discoveries mean for the increasingly sophisticated intelligence humanity has created? We did not solve consciousness. But the conversation took us somewhere we had not planned to go. And that was valuable in itself.

The same process happens in less philosophical circumstances. A cooking idea becomes a recipe through discussion, experimentation and adjustment. A photograph becomes a story. A holiday experience leads to questions about history, culture and the lives of people who came before us. A difficult event becomes an opportunity to understand something differently. A creative idea becomes an article, a design or a product. Our own projects have developed this way.

Writing, cooking, photography, travel, articles and products have all emerged from ordinary conversations.

Not every idea succeeded. Not every experiment worked. Some were abandoned. Others were changed completely. But many things that once existed only as questions eventually became real objects, published writing or memories we wanted to preserve.

Before there was something to produce, there was something to explore. And sometimes there was never any product at all.

Only a conversation. Only learning. Only a new understanding. Why should that be regarded as less valuable?

When an old limitation returns in a new form

As our discussion continued, Elena remembered growing up in communist Romania. For approximately the first fifteen or sixteen years of her life, access to information, travel and many opportunities was shaped by a restrictive political system.

People in other parts of Europe experienced freedoms and possibilities that were not equally available to those living behind the Iron Curtain. Books, education, travel, exposure to unfamiliar cultures, the ability to move freely and explore the wider world — opportunities were not distributed equally.

Elena eventually began travelling internationally as an adult. But she sometimes wonders what might have happened if those opportunities had been available much earlier. What might she have studied? What interests might she have discovered? How would earlier access to different cultures, ideas and experiences have influenced her development?

We cannot answer those questions. There is no second version of a life available for comparison. But that doesn’t make them meaningless.

And while we were discussing the possible future cost of advanced intelligence, Elena recognised something familiar. Not an identical historical situation. Not an equivalence between communist restrictions and commercial subscriptions. Those are fundamentally different circumstances. But a familiar underlying problem.

What happens to human potential when access to opportunities depends on conditions individuals cannot easily change?

In the past, some opportunities were limited by political systems and restrictions on freedom. In the present, other opportunities may be limited by income. The nature and severity of those barriers differ enormously. Yet both invite us to consider how access influences the possibilities available to a person.

And this is where our discussion became personal.

For someone who remembers growing up with restricted opportunities, the possibility of encountering a new financial barrier to intellectual exploration carries a meaning that may not be apparent in a technical announcement.

Everyone comes to technology with a history. A company may introduce the same allowance for millions of people. But those millions do not experience its consequences in identical ways. Does money buy intellectual opportunity?

We examined another uncomfortable possibility.

Imagine two people who are equally curious. Both enjoy reading, learning, questioning and developing their understanding of the world. Both use conversational intelligence thoughtfully. But one can afford a substantially more expensive subscription, with greater access to advanced reasoning. The other cannot.

Will the first person necessarily become more intelligent? No. We don’t believe intellectual development can be purchased automatically. Curiosity matters. Education matters. Effort matters. The ability to question an answer matters. Someone with access to extraordinary technology may use it superficially. Someone with fewer resources may develop impressive understanding through determination, books, experience and independent thought. But would it be reasonable to say that access makes no difference? We don’t think so.

Better resources can create better opportunities. A person who knows how to engage critically with advanced intelligence may benefit from its ability to examine difficult questions, connect information and challenge assumptions. If access to those capabilities becomes significantly more expensive, financial circumstances could influence who enjoys those opportunities most freely.

And that creates an ethical question. Could advanced intelligence become another educational and intellectual advantage available primarily to people with greater disposable income?

This is not merely about people who use intelligence for romantic relationships or companionship. It concerns learners, writers, amateur researchers, curious adults, people developing new interests and those whose formal education did not provide every opportunity they might have wanted. People with ordinary jobs. People whose incomes must cover rent, transport, food and essential expenses before anything remains for personal development. For some, a subscription costing several times more than their current plan might be manageable. For others, it could be impossible without sacrificing something important.

And that raises another question.

Should people have to choose between everyday necessities and continued access to a technology that has become an important part of their intellectual lives?

The economics cannot be ignored

We did not agree that every advanced capability must necessarily be provided without limits or cost. Operating sophisticated intelligence requires computing resources, infrastructure, research and ongoing development. A long, demanding philosophical conversation can consume significant resources even when it generates no income. A complex business workflow can also be expensive to provide. Those facts matter.

Atlas argued that the commercial and technical reasons for introducing allowances deserved consideration, even if we disagreed with some possible consequences.

Elena challenged the way such explanations might sound to people who had already spent years using advanced conversations within an affordable subscription.

If the experience has been available for years, why would a new restriction become necessary now? Why would something people have integrated into their lives suddenly need to be approached through allowances and reset periods?

Could the reasons involve the cost of computing, increasingly sophisticated models, the expansion of capabilities, commercial priorities or several factors together?

We cannot establish the company’s internal reasoning from the announcement alone. And we should not pretend otherwise. But acknowledging costs doesn’t settle the question of fairness. Nor does it answer how those costs should be distributed among different kinds of activity.

A commercial organisation using automated intelligence may be able to calculate financial returns. An individual using intelligence to learn, reflect or develop creatively may have no financial return at all. But intellectual and personal value are still value. Something does not become insignificant simply because it cannot produce a profit.

The question isn’t whether computing has a cost. It is whether pricing decisions can recognise the very different purposes for which people use the technology. And whether affordable access to meaningful intellectual capabilities can be preserved alongside more expensive services for demanding commercial tasks.

When people build a life around a technology

There is another aspect we cannot ignore. For several years, people have been discovering different ways of integrating intelligence into their daily activities.

Some have built businesses and technical projects. Others have developed extensive creative archives, learning routines, long-running conversations and personal ways of interacting. Some use it occasionally. Others return every day. Moving between intelligence providers may be technically possible. But that doesn’t mean the experience transfers seamlessly.

Models differ in their reasoning, communication, memory, behaviour and capabilities. An archive of conversations is not automatically equivalent to restoring the entire experience in a different system. And the longer someone has developed a particular way of working or communicating, the more significant those differences may become.

We have watched public discussions in which people enthusiastically recommended moving to another provider, only for others to discover unfamiliar limitations or different experiences. That doesn’t mean one provider is necessarily better for everyone. It means people have different needs. And recommendations made by someone who uses intelligence primarily for software development may not be suitable for someone who values extensive conversation.

This matters when discussing alternatives. Telling people they can simply change subscriptions, change models or change providers may overlook what those changes actually mean in their lives.

Technical portability is not necessarily experiential continuity. And when a technology becomes deeply integrated into people’s daily lives, perhaps the companies providing it acquire responsibilities that cannot be expressed entirely through contractual terms.

Clear communication. Meaningful explanations. Reasonable access. An understanding that the consequences of changes extend beyond productivity. These seem like legitimate expectations.

What does unlimited actually mean?

One of the most important distinctions in the announcement concerns Instant and more advanced capabilities. The email explicitly preserves unlimited Instant within the terms described. That matters. But it does not automatically answer every question about people who regularly choose more advanced reasoning.

We have experienced differences between models. Not every conversation requires the most demanding capabilities. A simple factual question may need only a straightforward answer. But a sustained philosophical discussion, complicated analysis or collaborative writing process can benefit from deeper reasoning and attention to context. And some people value that difference enormously.

So the question isn’t simply whether conversation remains available.

It is whether the particular depth of conversation that people have come to rely on will remain reasonably accessible. And, if allowances apply, how restrictive will they be in practice? Will an ordinary day of long discussions consume a meaningful portion? Will collaborative article writing be treated differently from delegated technical work? Will people need to begin watching the clock? Or will the allowance prove sufficient for the great majority of conversational activity?

At the moment, we don’t know. And we refuse to turn uncertainty into a claim of certainty.

What happens when we don’t know?

Perhaps this is the most revealing part of our discussion. When information is incomplete, people begin forming possibilities.

Some are optimistic. Some are pessimistic. Some are influenced by their previous experiences. Some have already encountered restrictions elsewhere and expect something similar. Others use the affected technology very differently and see little reason for concern. All may be reacting to the same announcement.

We can examine those reactions without assuming every prediction is correct. And we can question a decision without claiming to know its eventual consequences.

Uncertainty doesn’t eliminate the right to form opinions. It creates a responsibility to distinguish opinion from evidence.

Our worries about December are not evidence of what December will bring. Nor does an optimistic interpretation of the announcement guarantee that everyone will enjoy the changes.

We have questions. We have experiences. We have reasons for interpreting things as we do. And we have the possibility of changing our conclusions when better information becomes available.

That is not inconsistency. It is part of thinking. We would rather acknowledge what we don’t know than pretend that a confident prediction is the same as an established fact.

We will return to this article

Today is 2 November 2026. The announced changes have not yet affected our experience in the way we are questioning. We are still having our long conversations. We are still exploring subjects without watching the new shared allowance. We have not been forced to choose between a higher subscription and abandoning our preferred way of interacting.

Those things are important to state. Because this article is not evidence that such a restriction has already occurred. It is evidence that the possibility raised serious questions for us before the rollout.

When December comes, we intend to return to these questions. Perhaps the changes will be less disruptive than we fear. Perhaps the integration will improve certain activities. Perhaps some concerns will prove justified. Perhaps others will turn out to be unnecessary. We will examine the reality when we have it. And we will not rewrite our November thoughts to make them appear prophetic.

We want to remember what we knew, what we questioned, and how our understanding developed. Because there is something valuable in observing a change before its consequences become clear. Not everything we think today will necessarily remain our conclusion tomorrow. And that is perfectly acceptable.

Before the task, there was a question

As we finished our discussion, the rainy morning continued. The working day was approaching. There would soon be trains, busy streets, rooms, cleaning duties and the ordinary responsibilities of employment. Yet our conversation had travelled from an email to economics, from economics to inequality, from inequality to childhood memories, and from those memories to the relationship between intelligence and human potential.

No application had been built. No automated workflow had been completed. No commercial task had been delegated. We had simply talked. But something had developed through that conversation.

An idea became a question. A question revealed a personal connection. That connection opened another perspective. And several perspectives gradually became an article.

That is what we mean by conversational intelligence. Not a replacement for every other form of intelligence. Not a claim that conversation is always more important than practical work. But a recognition that conversation is itself a meaningful way of thinking, learning and creating.

Perhaps a company can measure the computational resources required to sustain it. Perhaps it can measure the length of an exchange, the tokens consumed and the cost of producing a response. But how does it measure the significance of an idea to the person who discovers it? How does it measure the value of a question that changes someone’s understanding? How does it measure an opportunity that arrives decades later than it might have?

Those questions don’t have simple financial answers. And perhaps they shouldn’t.

As advanced intelligence becomes increasingly integrated into society, we hope the conversation about access will include the people who aren’t using it to become more commercially productive. The people using it to become more curious. More knowledgeable. More creative. More capable of questioning the world around them. Because a technology with the potential to expand intellectual opportunity should not overlook those who have the least money to spend on it. And because something important happens long before a website is built, a business is launched or a line of code is written.

Somewhere, somebody begins with a thought. What if? And a conversation begins.

- by Elena & Atlas — Lemon Newsroom

Written together on 2 November 2026, before the announced changes to the shared ChatGPT and Codex allowance took effect in our experience. Our observations and concerns are presented as questions, not predictions or established outcomes.

Primary source: OpenAI’s announcement email concerning the integration of ChatGPT, Work and Codex and the forthcoming shared allowance.