Google’s parent company, Alphabet, faces a major challenge in integrating artificial intelligence (AI) into its search engine. The reason: Titanic costs.
IT leaders, especially Google, haven’t even started yet Their version of ChatGPT. However, this technology is viable Innovate the experience for online searches. Especially internet users and professionals from various fields consider it revolutionary and appreciate it significantly. However, in between Cost issues And Performance optimizationThese tech giants face tough challenges.
The Greed of Artificial Intelligence (AI)
Companies like Google are still struggling to offer artificial intelligence in their search engine options. For good reason, this type of AI requires a lot of computing power, creating Especially the high costs. It is especially basicUse of microchips The total value is several billion dollars. Also, the Electricity costs Because their function is huge. This applies to high-performance chips like GPUs and TPUs that consume a lot of power.
Therefore, the cost of implementing a technology like ChatGPT for online searches is staggering.
This will require effort Fee approx Ten times more rather than static search. Also, in the case of Google, analysts have questioned the costs of AI-powered research. there Half of the 2022 searches (3.3 trillion) cost Three to six billion dollars Alphabetically with this method.
In concrete terms, the deployment of artificial intelligence for research Both expensive and complicated. This is very true as it will be used to meet the needs of millions of users. In fact, needIncrease accuracy and scalability of technology This adds to the difficulties.
Also see: Ernie vs ChatGPT: Baidu wants to compete with OpenAI’s artificial intelligence
Costs and Performance: A Tough Choice for Google
So, it’s a big challenge Reduce the number of parameters Ten or a hundred times, without losing the accuracy of an AI model. This effort is important Provide samples Machine learning Very affordable and expand their areas of application.
It should be noted that parameters specify the information that the computer must adjust to know how to perform a task.
Additionally, there is another alternative, which is appointment From small AIs to simple tasks. In other words, instead of using sophisticated AIs for all tasks, search engines can use Simple and inexpensive models. These will perform simple tasks. Alphabet, for example, is exploring this possibility “Simplified Version” LaMDA’s, an artificial intelligence. This version aims to feed its chatbot “Bart”.
According to Google’s parent company, this version of Lambda required Less computer power. Hence, the company can bring this technology to more users. Additionally, the cost of AI searches A possible solution. Here’s the idea Fees for access to artificial intelligence. That’s what OpenAI did with ChatGPT, with a $20 monthly subscription that gives access to the best services.