I recently came across a blog post penned by Sam Altman, the CEO of OpenAI, detailing his three significant insights into the accelerated development of general artificial intelligence. Notably, he hinted at an expected decrease in the cost of AI that could potentially drop tenfold each year.
According to Altman, the cost of utilizing a certain level of AI decreases approximately tenfold every year, and as prices decrease, there is significantly increased usage. For instance, the price per token for GPT-4 dropped around 150 times from its early 2023 cost to mid-2024 when it became GPT-4o. This rate of change is far greater than Moore’s law, which only saw a doubling every 18 months.
In the detailed and broad blog post, Sam Altman listed these key highlights:
- The intelligence of an AI model roughly equals the log of the resources used to train and run it.
- The cost to use a given level of AI falls about 10x every 12 months, and lower prices lead to much more use.
- The socioeconomic value of linearly increasing intelligence is super-exponential in nature.
In simpler terms, Altman’s main findings appear to focus on the latest advancements in the field of artificial intelligence. For instance, he highlights the appearance of DeepSeek, a more affordable alternative to ChatGPT Pro from OpenAI which costs $200. Additionally, he mentions that recent trends suggest that scaling laws might be slowing down the growth and evolution of advanced AI models.
DeepSeek stands out distinctly among AI industry players, demonstrating that startups don’t always need substantial financial resources to make an impact in the AI field.
Initially, it was reported that the research team associated with the submission stated that their company invested just $6 million in training their AI model. However, more recent news has cast a negative light on this Chinese startup following accusations that they illegally used copyrighted material from Microsoft and OpenAI to train their so-called “ultra-cost-effective” AI. Additionally, it was alleged in another report that the company spent an estimated $1.6 billion and utilized 50,000 NVIDIA AI chips in the development of their AI model.
Although the model has been widely praised, primarily due to its open-source nature, it experienced a significant “large-scale cyberattack” that led the AI company to temporarily suspend new user registrations. Sam Altman posits that the economic value derived from linearly increasing intelligence is exponentially magnified. He further stated, “Given this, we don’t foresee the need for exponentially growing investment to slow down in the immediate future,” concluding his thoughts.
It’s troubling to note that a harmful software was found on DeepSeek’s website, which reportedly shared user data with the Chinese government without proper approval, even though there is a U.S. ban in place.
In essence, Altman emphasizes that an AI model’s intelligence is heavily dependent on the resources allocated for its training and functioning. More specifically, these resources include computational power for training, data, and computational power for inference. It seems that by investing more money, you can expect consistent and reliable improvements; the scaling laws predicting this have been found accurate across a wide range of magnitudes.
Altman’s comments follow OpenAI’s revelation, several weeks ago, about their $500 billion investment in a project called Stargate, which aims to build data centers nationwide in the U.S., thereby strengthening the progress of AI technology.
In the previous year, it was suggested that leading AI research facilities like OpenAI, Google, and Anthropic struggled to create sophisticated AI models due to a scarcity of top-tier training material. Yet, Yann LeCun and ex-Google CEO Eric Schmidt contested this notion, asserting there’s no concrete proof suggesting scaling laws are hindering the advancement of AI technology. Altman further emphasized, “There is no barrier.
Read More
- POL PREDICTION. POL cryptocurrency
- INJ PREDICTION. INJ cryptocurrency
- BONE PREDICTION. BONE cryptocurrency
- DUSK PREDICTION. DUSK cryptocurrency
- DGB PREDICTION. DGB cryptocurrency
- HBAR PREDICTION. HBAR cryptocurrency
- FIL PREDICTION. FIL cryptocurrency
- LDO PREDICTION. LDO cryptocurrency
- ILV PREDICTION. ILV cryptocurrency
- Blockchain is the best fintech to ensure Sharia ethics — Web3 exec
2025-02-11 15:16