From plateau predictions to buggy rollouts — Bill Gates’ GPT-5 skepticism looks strangely accurate

by

Prior to its release, there was much excitement and high expectations for GPT-5. OpenAI advertised the AI model as the most intelligent, likening it to a team of PhD experts. Upon its launch, GPT-5 came equipped with an array of cutting-edge capabilities spanning numerous areas such as coding, writing, and medicine.

The leader of the company that created ChatGPT, Sam Altman, once predicted that technology more intelligent than the smartest person you can think of will soon fit in your pocket, which might refer to GPT-5. However, after its release and the decision to retire its earlier versions, the AI firm has faced criticism from users.

The user expressed concern and disappointment, stating that “ChatGPT seems to have lost its charm” due to recurring issues such as bugs, glitches, and unresponsiveness. However, recently, Altman has announced some updates for the release of GPT-5. These include an increase in rate limits for ChatGPT Plus users, continued access to GPT-40 for Plus users, and a commitment to providing more clarity about which model is responding to each query.

1. For ChatGPT Plus users, we’re increasing GPT-5 rate limits by double as the rollout is completed.
2. We grant Plus users the option to continue using the current model, while monitoring usage and considering how long to maintain legacy models.
3. Starting from August 8, 2025, GPT-5 is expected to appear more intelligent.

During its initial release, the model struggled due to a malfunctioning feature in GPT-5 known as the autoswitcher, which made it appear less intelligent. OpenAI alleges that they have since addressed this issue, but two years ago, Microsoft co-founder Bill Gates may have foreseen such problems with GPT-5, even before it was launched (as reported by The Indian Express).

The model’s poor performance upon launch can be attributed to a faulty feature in GPT-5 called the autoswitcher, causing it to seem less intelligent. OpenAI maintains that they have resolved this issue, yet two years ago, Bill Gates of Microsoft may have anticipated challenges for GPT-5 even before its creation, according to The Indian Express.

Sam Altman earlier asserted with a significant level of confidence that GPT-5 would outsmart GPT-4. However, he acknowledged that GPT-4, despite its flaws, was far from satisfactory and at best a source of minor discomfort.

In an interview with Handelsblatt, a German business newspaper, in October, Bill Gates expressed that the development of GPT technology had plateaued, contrary to the general belief among OpenAI’s team members, including Sam Altman, who anticipate that GPT-5 will outperform GPT-4 substantially.

Although the philanthropist billionaire finds the advancement from GPT-2 to GPT-4 astonishing, he expressed a degree of doubt, inquiring if OpenAI could deliver comparable outcomes when moving on to GPT-5.

In line with his statements, many users have voiced their discontent with GPT-5, likening it to only a marginal upgrade compared to GPT-4. Gatesa anticipated that advancements in the generative AI terrain would slow down post GPT-4, implying that OpenAI may have reached its peak in terms of evolving its GPT technology.

Despite the latest findings, AI has the potential to reach greater levels of sophistication, enhancing its dependability and providing enhanced healthcare guidance through smartphones. Notably, AI necessitates substantial financial investments and computational resources for operation.

Indeed, training a large language model can be quite costly. Initially, the cost per query was about ten cents, but today it’s likely closer to three cents. However, the expenses associated with computational power or semiconductors continue to be massive.

Reports of OpenAI hitting a wall in AI development are not new

In the past year, a report surfaced suggesting leading AI research institutions such as OpenAI, Google, and Anthropic were encountering difficulties in creating sophisticated AI models. These challenges are believed to stem from insufficient high-quality data for training and the expensive pursuit of AI trends. The report also suggested that the delays in introducing new-generation models are closely linked to these problems.

Alternatively,

Sam Altman, OpenAI’s CEO, swiftly dismissed the claims, stating that “there is no wall” when it comes to the AI firm’s ability to train models, implying they haven’t yet hit a knowledge limit. Eric Schmidt, a former Google CEO, appeared to share similar views, asserting that there’s currently no proof that growth limitations have started setting in.

Over the next five years, you can expect to see improvements of two or three iterations in these massive models, as they are advancing at an unparalleled rate. So far, there’s no indication that this rapid advancement is slowing down. However, it will eventually level off, but we’re not quite there just yet.

I wonder if the upcoming version of OpenAI’s GPT, namely GPT-5, will meet expectations and tackle the concerns raised by some users.”

In response to your question, “What are your thoughts on OpenAI’s GPT technology plateauing?”, I would say that it’s important for OpenAI to continue innovating and addressing user feedback in order to maintain the success of their GPT technology. However, only time will tell if the next iteration of GPT lives up to the hype. Let’s see what improvements OpenAI brings with GPT-5 and how well they address the issues highlighted by users. Feel free to share your thoughts in the comments!

Read More

2025-08-16 17:15