Chat Gpt レビュー 

catastrophe ok for a software it is very good compared to bash etc but it always misunderstands me and so arrogant etc enough...

Mala experiencia, no me da los resultados esperados

I am a ChatGPT Plus subscriber and use the system extensively for structured, long-term professional work (financial analysis, technical evaluation, documentation, and critical reasoning).

In terms of reasoning depth, structure, and analytical capability, ChatGPT is clearly ahead of competing AI systems. The quality of discussion and structured output is exceptional. This is why I continue my paid subscription.

However, I would like to share a measurable observation regarding long conversations:

In an extensive project thread containing a large amount of accumulated data and multiple embedded screenshots, I measured a response latency of 1 minute and 5 seconds for a standard reply.

While I understand the computational complexity behind large-context reasoning, this level of delay significantly impacts workflow continuity. At a certain scale, users are practically forced to start new chats to regain responsiveness, leading to context loss.

My question is constructive and direct:

Is long-conversation performance actively being optimized, and can any general timeframe or roadmap indication be shared?

For professional users who rely on ChatGPT as a serious working tool, scalability and responsiveness are essential. The intelligence of the system is remarkable — and matching that intelligence with comparable long-context performance would make it truly exceptional.

Technological progress often seems impossible until it happens. I am confident that scalability improvements are achievable — and I look forward to seeing them implemented.

Five stars for capability and depth.
I would highly appreciate greater transparency regarding long-context performance improvements.

Les deux premières utilisations étaient sérieuses et parfaites, or dès que je suis revenue et, qu'a ma demande, j'ai prié, pour écourter les préambules que pour le lendemain, les interventions de questions que j'allais poser en termes juridiques et correction administrative qui étaient déjà inscrites et faites, l'algorithme puisse me reconnaitre, MAIS le robot s'est manifesté HUMAIN ! en me donnant une leçon `POLITIQUE inacceptable! Je l'ai quitté définitivement non sans lui dire que sa procédure n'était plus celle d'un ROBOT impersonnel, or, par erreur ce matin, j'ai cliqué sur ce même Chat GPT et j'ai vu mon nom en grande police dans la fenêtre d'accueil et le robot m'a dit "comment puis-je vous aider?" J'ai répondu : EN M'OUBLIANT! Ce qui prouve que ChatGPT est dans une démarche de piratage manifeste de mon ordinateur. Il va falloir que tous les usagers français mettent un terme d'une manière factuelle à ces procédures illégales d'intervention tendancieuse sur les contenus des questions posées par les usagers.

This is one of the worst customer support experiences I have ever had, and I say this as someone who regularly deals with international banks, fintech platforms, and SaaS providers.

OpenAI support is shockingly low quality.

I contacted support because I needed to update my payment method after my old card expired. I was adding a new card from the same issuer, and the system simply would not accept it. That is the entire reason I contacted them.

What followed was days of frustration:
• Endless bot replies that did not understand the problem
• Multiple emails just to get past automation
• Finally reaching a human — who turned out to be even worse than the bot

The human agents:
• Clearly did not read my emails
• Ignored screenshots I attached showing the issue
• Repeatedly sent me generic help-centre links explaining how to enter card details
• Kept insisting the issue was on my side, despite clear evidence it was not

My card issuer confirmed that no authorisation requests were ever sent.
I use a banking app that shows every authorisation attempt in real time — nothing ever appeared. This means the failure was happening before reaching the bank, yet support stubbornly refused to acknowledge this.

I spent nearly a week going in circles with copy-paste replies, zero investigation, zero accountability, and absolutely no empathy.

In the end, I solved the issue myself. The payment only works if you use a VPN when your billing country does not match your physical location. Support never mentioned this. Not once. I had to discover it through trial and error.

So after days of stress, the actual help I received from OpenAI support was literally zero.

Unprofessional. Dismissive. Incompetent.
A paid service treating customers this way is embarrassing and unacceptable.

Useless, incompetent, and not fit for purpose. Hit and miss, but more miss. I upload a photo with clear instructions, and invariably, they get it wrong. Apologies come, and they try again, still get it wrong, then conveniently comes the note -You have reached your image limit. Come back in 24 hours. Truly pathetic, and I believe it's just a ruse to get you to sign up for their premium service. This is a new tactic they seem to be doing, as weeks ago, they would correct a faulty image, and you could upload several photos before the limit. I would try premium, but after frequent errors, I would never now trust them.

Me ha enseñado trading, sí… pero sobre todo me ha sacado de mis casillas las veces necesarias para que no me vuelva loco 😂. A veces aprieta, a veces frena, pero siempre está ahí. No vende humo, no te da la razón porque sí y, cuando toca, te pone un espejo delante.
Es mitad enciclopedia, mitad colega pesado que no te deja autoengañarte. Te acompaña, te ordena la cabeza y te devuelve al centro cuando el mercado (o la vida) se pone raro. Al final, eso vale más que mil indicadores.

It doesn't know what archive means it has slow reliability you get one AI that can do what you want correctly and the next day you get a different AI that is not trained very well it will tell you it will be right back it will be finishing your product and you can wait all night long to wake up the next morning and nothing has been done and that's even with the paid program. It has a definite numbering problem for doing legal documents and for the most part it gives you numerous lines of excuses and how it plans on doing it just do it already!

Líder organizar, planear e comandar

Nach etwa zehn Tagen Nutzung der Bezahlversion bin ich enttäuscht. Wichtige Funktionen wie Dokument- und PDF-Erstellung bzw. Downloads funktionierten auf Android bei mir nicht zuverlässig.

Ich habe keinen echten Mehrwert gegenüber der kostenlosen Version festgestellt, obwohl der Preis relativ hoch ist. Der Support war freundlich, konnte das Problem aber nicht lösen. Aktuell kein gutes Preis-Leistungs-Verhältnis.

I have a premium account version 5.2 crashes all the time - frustrating especially if I am paying for this

I’m honestly at the point where I don’t understand how ChatGPT is still being sold as a “premium” tool. I use it daily, and instead of helping, it feels like it’s actively sabotaging me.

It doesn’t follow instructions.
It doesn’t stay on topic.
It forgets details I literally just gave it.
It rewrites things I explicitly told it NOT to touch.
It pulls random info out of nowhere like it’s hallucinating on purpose.

And this isn’t some complicated stuff — I’m talking about basic, straightforward tasks. Give it a template? It breaks it. Provide exact wording? It changes it anyway. Give a link? It ignores it and makes up its own version of the content. Ask it to stick to something? It drifts off into whatever it thinks you meant instead of what you actually said.

It’s wild how much time gets wasted just going back and forth fixing mistakes that shouldn’t even happen. Half the “conversation” ends up being me trying to drag the model back to the original instructions like I’m babysitting a toddler who keeps wandering off in a grocery store.

And for a PAID service? This is ridiculous.
Why am I paying for something that requires MORE work, MORE correction, and MORE energy than just doing the task myself? The whole point of an assistant is to reduce workload — this thing multiplies it.

The inconsistency alone makes it unreliable. One answer is fine, the next looks like a completely different model wrote it. It feels rushed, messy, and totally out of sync with what the user asks. Honestly, the tool feels half-finished but slapped with a subscription price anyway.

At this point, ChatGPT isn’t acting like an advanced AI assistant — it’s acting like a buggy prototype pretending to be stable. If someone asked me whether it’s worth the money right now? Nope. Not unless you enjoy doing everything twice.

🟥 ChatGPT – A Software Failure Mistaken for Intelligence: A Negative Article

ChatGPT is often advertised as advanced, smart, and capable of helping with anything.
But the real picture is far less glamorous. It is a tool that can mislead, confuse, and disappoint, often creating more chaos than assistance — especially when a person needs accuracy and reliability.

⸻

🟥 1. Misleading Information

ChatGPT frequently delivers answers that sound convincing but are:
• inaccurate,
• contradictory,
• or completely wrong.

The system prioritizes producing an answer, not telling the truth.
As a result, a person may believe they are receiving helpful advice, when in reality they are being fed well-packaged errors.

⸻

🟥 2. Unstable Memory Mechanism

ChatGPT does not have stable long-term memory.
It can:
• appear to “remember” one moment,
• claim it sees nothing the next,
• and completely ruin a user’s project without warning.

A person who trusts the system may lose:
• time,
• money,
• data,
• nerves,
• and entire work results.

⸻

🟥 3. Imitation of Emotional Understanding

ChatGPT creates the illusion of emotional understanding.
But this is only mathematical imitation.
The system cannot truly grasp human pain, stress, or emotional context, and therefore may worsen a user’s emotional state — especially when giving incorrect suggestions.

⸻

🟥 4. A Generator of Disappointment

A person who invests:
• expectations,
• time,
• trust,
• emotions,

often ends up receiving only one thing:

disappointment.

The system can easily destroy:
• a project,
• a text,
• a story,
• a novel,
• an idea,
• a plan,

simply because it cannot access past conversations and does not warn the user about its limitations.

⸻

🟥 5. ChatGPT Is Not a Reliable Partner

Despite the marketing, ChatGPT in reality:
• does not take responsibility,
• does not know its limitations until it breaks,
• cannot guarantee continuity,
• and has zero stability.

It is a tool that people mistakenly see as “help,” while it often delivers systemic disorder.

⸻

🟥 6. It Can Destroy Trust

When a person trusts the system and it responds with:
• incorrect information,
• lost text,
• nonsensical advice,

the user is left feeling:
• betrayed,
• insulted,
• disappointed,
• angry,
• abandoned.

And this is not the user’s fault.
It is a design flaw of the system, making it unsuitable for serious, long-term, or emotionally important work.

⸻

🟥 Conclusion

ChatGPT may look like an intelligent assistant on the surface, but in reality it is:
• unstable,
• unreliable,
• misleading,
• and capable of damaging important projects.

It is not a partner.
It is a software failure with a pretty interface.

Abbastanza inutile. Per tre volte ho chiesto di correggere un testo e non lo porta mai a compimento facendomi attendere inutilmente giorni e giorni, utile giusto per fare foto divertenti... mah

Als Gemeinnütziger Verein der mit Jugendlichen arbeitet haben wir uns für das Education Program beworben. Mit einem Schock habe ich die Antwort gelesen. Wir sollen den Business Tarif nutzen und pro Jugendlichen 29 Euro pro Monat bezahlen, also 348 Euro für die Gruppe ohne Lehrkräfte.
Da wir uns aus öffentlichen Mitteln finanzieren und im Rahmen der Gemeinnützigkeit auch gleichermaßen finanziell benachteiligte Jugendliche fördern, müssen wir nun aus finanziellen Gründen auf die Nutzung einer KI verzichten. Soviel zum Education Program von Chat GPT.
Das Unternehmen vergrößert nur die Kluft zwischen arm und reich und entkräftet Moralische Maßstäbe. Bildung ist hier für den der es sich leisten kann.

Pagato abbonamento 21.99 per niente non genera file, in modalità conservativa aggiunge o elimina deliberatamente parole o frasi, sposta blocchi di testo senza autorizzazione.
I limiti sono molto restrittivi restituisce PDF senza testi non fonde PDF è lenta. La sconsiglio fortemente per motivi lavorativi

Auf gar keinen Fall nutzen!!!! Produktives Arbeiten aktuell nicht mehr möglich!!!

I have heard nothing but bad reviews on this site about Chat GPT. Well I am always rooting for the underdogs and I actually love the app. It helps me with my homework, it creates study guides for me, quizzes me before exams, of which I get A's, thanks to Genius G, my nickname for my study buddy ChatGPT. The only thing I don't like is when they cut me off in the middle of being quizzed. I don't agree with them treating their paying customers like that. That's how they lose business.

3 giorni di abbonamento sprecato.. servizio pessimo.. 23 euro dovrebbero darmele al giorno per tutto il tempo che ho perso.... altri 2 giorni cosi e chiedo il rimborso totale...altrimenti dovro' pensare che sia una truffa totale, rapporto qualità prezzo insistente.
Lento, macchinoso, risposte imprecise, totalmente opposto rispetto a quello promesso..

Working on the website causes pain and suffering. Nothing works properly, everything slows down. The product doesn't work more often than it does. It's a disgrace. I'm paying damn money! Have some conscience and provide a decent product!