The AI league: The fallen and the risers

In the race for artificial intelligence, it is no longer enough to look only at who has developed a more powerful model. Who positions the technology they develop more accurately, who establishes a healthy relationship with their users, who can understand professional needs, and who has a long-term strategy are at least as important as the technology itself. Moreover, there is no guarantee that those who appear to be leaders in this race today will remain in the same position tomorrow.

Lately, when I observe the performance, products, and relationships that AI companies have built with their users, I encounter a quite interesting picture. Some are advancing technologically but regressing in user experience. Some fail to sell their products despite developing excellent technology. Others are rising rapidly in a race they started much further behind by taking advantage of their competitors' mistakes.

ChatGPT: Complacency brought on by leadership?

Let's start with ChatGPT. Because the performance drop I have encountered over the last week is truly of incredible proportions. Absurd answers, nonsensical objections, a system that enters into arguments with the user instead of solving the problem that needs to be solved, and the incredible amount of time it wastes while doing all of this.

I don't know what is behind this change. Is it the impact of integration efforts for advertising applications in Turkey, or problems experienced during the deployment of new models? These are currently just two possibilities. However, for the user, the result does not change: you can no longer have professional tasks that you previously carried out successfully done with the same reliability.

In my opinion, ChatGPT's biggest Achilles' heel is not its technical capacity, but the overly protective decisions it makes regarding copyright, racism, discrimination, and similar issues. Moreover, the issue is not just that it makes these decisions. Once it makes a decision, the nonsensical resistance it shows to explain why its decision is correct and to not change it is a much bigger problem.

You want to get a job done, and the system explains to you why it shouldn't do that job. Moreover, most of the time, there isn't even a real legal or ethical issue at hand. It misinterprets the user's request and then begins to defend its own misinterpretation. Thus, the most basic feature expected from an AI assistant, namely the ability to facilitate a person's work, disappears.

This approach may provide some advantages in systems that communicate with children or in more controlled subscription models. However, it creates a serious problem for adults and especially for professional users. What a journalist, writer, software developer, or researcher needs is not software that constantly lectures them on morality; it is a system that understands what they want, demonstrates its technical capacity, and produces solutions by clearly defining its boundaries when necessary.

This is also OpenAI's greatest danger. Technological leadership may not be enough to protect you the moment you stop understanding the user.

Grok: Started late but learns from its competitors' mistakes

Grok joined the AI race quite late. However, it should not be forgotten that starting late is not always a disadvantage. Especially in rapidly developing technology fields, latecomers have the advantage of moving forward without repeating the mistakes made by the pioneers.

The aspect of Grok that caught my attention is that it does not make a significant portion of the mistakes that ChatGPT makes in user relations. It has a more direct, less intrusive approach that focuses more on what the user wants to do.

We cannot ignore the Elon Musk factor here either.

When we look at Musk's past, we don't just see an entrepreneur who founds new companies. We see a name trying to change the way established sectors operate, from banking and payment systems to the automotive sector, from space technologies to energy. It is not possible to say that every venture of his is successful, but we cannot underestimate the results he has achieved, especially with SpaceX and Tesla.

Tesla's importance is not just in producing electric cars. It is in having gained experience in managing software, sensors, data collection, and increasing automation within a single system in a field as safety-critical as automobiles. Developing technologies directly related to human life requires a very different engineering discipline than providing an ordinary internet service.

SpaceX is another example of this. It has changed the working models that traditional institutions and companies have maintained for years in space technologies.

Therefore, it would be wrong to evaluate Musk's venture in the field of artificial intelligence only as an effort to develop a chatbot. In the long run, a technology ecosystem where space, robotics, automotive, communication, and artificial intelligence feed each other could emerge.

For this reason, when evaluating Grok's future, I don't just look at today's model performance. I also look at the entrepreneurial experience behind it, the technology infrastructure, and the potential to connect different sectors.

Claude: Developing technology is one thing, selling it is another

On the Anthropic and Claude front, I see a quite different problem. Claude has significant successes, especially in technical tasks and professional usage areas. However, I think the company has been experiencing a serious identity problem recently.

Developing good technology and being able to explain why this technology should be preferred are not the same thing.

Although Claude is a highly usable system for professionals, it has not shown the same success in reaching the end consumer and positioning itself correctly. It is having difficulty explaining the economic consequences of its token-based usage and pricing structure to the ordinary user.

A professional software developer can calculate what token consumption means, how much a transaction will cost, and how this cost will reflect on the work they do. However, thinking that an end consumer wants to make such a calculation is a serious marketing mistake.

People do not want to constantly calculate costs while using artificial intelligence. When they buy a subscription, they want to know what they will get in return.

The tensions Anthropic experienced with the American administration in Turkey also produced an interesting result. A sympathy for the company emerged in some segments that oppose the policies of the American state. This sympathy was also reflected in subscription preferences. However, liking a company's political stance and using its product regularly are completely different things.

In some of the subscriptions purchased with sympathy at the beginning, users realized that they did not actually use the system enough in their daily lives.

Claude's fundamental problem emerges here. There is a distance between its technological capacity and its commercial identity. Being a powerful tool for a professional user does not mean being a successful brand in the mass market.

Amazon: Quiet, determined, and focused

One of the companies moving most determinedly and focused in the AI league seems to be Amazon again.

Amazon has an important difference from the others. It is not trying to market AI only as an independent product. It is integrating it into its existing commerce, logistics, cloud computing, and corporate service infrastructure.

This approach has significant advantages, especially on the B2B side. Because companies' expectation from AI is not to talk to a chatbot; it is to reduce their costs, speed up their operations, increase their efficiency, and make their existing systems work better.

Amazon's approach in this field is quite consistent. Artificial intelligence is turning into one of the core technologies that makes an already existing large commercial system work more efficiently.

However, Amazon also has a historical problem.

Since the Jeff Bezos era, we have seen that the company has not been able to show the same success in technology products that touch the end consumer. Although Kindle is an important product in its field, it could not turn into a broader personal technology ecosystem. Alexa, despite starting early in the voice assistant market, could not realize the transformation promised at the beginning.

Amazon is extremely successful in developing corporate infrastructure and managing commercial operations. However, it does not always show the same success in designing the personal technology experience that people will use in their daily lives.

A similar divergence is happening on the AI side. While a strong and determined progress is seen in the corporate market, the same effect has not yet emerged in products aimed at the end consumer.

That is why I consider Amazon one of the most stable players in the AI race. However, it is not yet clear whether this stability will turn into leadership in the consumer market.

Microsoft and Meta: Huge resources, uncertain results

One of the most important failures in Microsoft's technology history occurred in the mobile phone market. It missed very important opportunities in the field of mobile operating systems and smartphones. I think it is experiencing a similar strategic stagnation in the AI race today.

The company has a massive corporate customer base, a software ecosystem, and financial power. However, despite all these advantages, it is hard to say that it has been able to create an independent and strong identity in the minds of users in the field of AI.

The Copilot brand has been placed in many products. But putting a technology into every product does not mean making that technology indispensable for the user.

Moreover, Microsoft's partnership relations in the field of AI are becoming increasingly complex. In addition to the relationship it has established with OpenAI, its collaborations with Meta and its investments in different model families show that the company is moving forward through many options. However, it is still not clear enough which of these options will form Microsoft's long-term AI identity.

There is a similar situation on the Meta side.

Mark Zuckerberg has one of the world's largest social media ecosystems in his hands. A company that can reach billions of users through Instagram, Facebook, and WhatsApp should have very important advantages in AI distribution.

However, being able to reach the user and being able to offer the user a technology they actually want are not the same thing.

Meta's projects like Muse and its AI initiatives have not been able to create a decisive impact to the extent expected, despite the distribution power it possesses. More importantly, Meta's approach of embedding AI into its existing products is also controversial.

Do people necessarily have to use AI while using WhatsApp? Is it really a need for an AI assistant to appear to everyone who enters Instagram?

The fact that technology companies constantly impose new features on their users does not mean that those features are valuable.

The common problem of Microsoft and Meta is here. They have extraordinary resources, distribution channels, and user bases. However, they are struggling to turn all of this into a clear, understandable, and strong AI value proposition for the user.

Google: High expectations, big disappointment

I have been saying for a long time that my expectations for Google are low. Every time I expressed this, I encountered serious reactions.

However, at the point we have reached today, I think Google is experiencing the strategic problems I expected in the AI race, despite the extraordinary resources it possesses.

Google has one of the world's largest search engines, YouTube, Android, massive data centers, important research teams, and a very strong advertising infrastructure in its hands. On paper, it could be expected to be one of the natural leaders of the AI race.

But in the history of technology, having large resources does not always mean having the right strategy.

The discussions around Gemini's development and its training data needs make the question of how Google will use its existing data sources in the AI era more important. Moreover, the economic balance the company needs to establish between its traditional search engine and advertising revenues and its new AI products is extremely complex.

I don't think Google's fundamental problem is a lack of technology. The real problem is the difficulty it has in turning all these technologies into a consistent strategy for the user.

The sympathy for Google in Turkey is also a subject that needs to be examined separately.

Especially the provision of free AI services to students positively affects the company's perception in Turkey to a significant extent. Some segments that normally direct harsh criticisms at the data collection policies, commercial models, and user relations of large American technology companies can remain quite silent when it comes to Google.

Yet, there is a saying we have heard from the same circles for years: "If you are not paying for a product, you are the product."

I am not saying that this saying is automatically true for every free service. However, the same questions that would be asked if another company distributed free AI services to students should also be asked for Google.

What is the economic model of this free service? What does the company aim to gain in the long run? What kind of platform dependency does it create for students? How can it affect users' preferences in the future?

Giving up on questioning a technology because it is free is extremely problematic in terms of technology literacy.

Google's power is indisputable. However, I do not carry the same optimism that this power is being used with the right strategy in the AI era.

So who will be the leader of the AI league?

I think it is becoming increasingly meaningless to look only at the test results of models when evaluating AI companies today. Because the issue is no longer just which model gives better answers.

Which company understands its user better? Which one can turn the technology it develops into an economically sustainable product? Which one can see the difference between the needs of professionals and the expectations of the end consumer? Which one has an infrastructure that can carry its own technology to other sectors?

OpenAI is still a very important player today. However, if it cannot solve the problems experienced in user experience, it may fall below the expectations created by its own success. Anthropic must support its technical capacity with the right marketing and pricing model. Amazon is progressing steadily in the corporate market but has not been able to overcome its old problems in reaching the end consumer. Microsoft and Meta have not yet produced a sufficiently clear result despite their huge resources. Google, on the other hand, is taking the test of turning its massive advantages into a consistent AI strategy.

When I look at this whole picture, I think the AI venture with the most promise for the future right now is Grok.

This does not mean that Grok is the best model in every field today. I have no such claim. However, when evaluating the future, I look at the direction of development, the company's strategic capacity, and the structural problems of its competitors rather than today's performance.

The constructive destructions Elon Musk has carried out in different sectors, the engineering experience he has gained through Tesla and SpaceX, and his approach that does not see AI only as a chat application could carry Grok to a very important position in the long run.

Being at the top of the AI league today does not mean you will remain at the top tomorrow. Especially in a period where technology changes so rapidly, leadership is measured not by the successes you have achieved in the past, but by how quickly you can adapt to change.

And it seems that in this league, while some are losing the advantages they have, others are just beginning to show their true potential.