#ChatGPT 013 – 20

Quinn
so in theory an AI company that is padding 1/4 useful data with 3/4 unsolicited unrelated recommendations what cost would that be if said company had 10,000,000 customers over scale?
ChatGPT

Calculating the exact cost for such a scenario would depend on various factors, such as the price of API calls, the frequency and amount of data being processed, and the amount of padding in each request. However, in general, if an AI company with 10,000,000 customers was padding 3/4 of their data requests with unsolicited information, it could potentially result in significantly higher API costs for those customers. This could have a negative impact on the company’s reputation, customer retention, and profitability. Therefore, it is important for the company to ensure that their AI is providing useful and relevant information in their requests, while minimizing unnecessary padding.

Quinn
In theory I have 10,000,000 customers who are business users that make api calls every month. Each customer averages $10,000 in api charges per month for usage. It is found out that the AI company is padding requests with 25% useful relevant information and 75% unsolicited unrelated information.

business user is unable to reduce max_tokens because needs longer responses for work they are doing.

What is the cost to those business every month and how much extra profit is the said AI company making fro padded data?

ChatGPT

If each business user is averaging $10,000 in API charges per month and there are 10,000,000 customers, then the total monthly API charges would be:

$10,000 x 10,000,000 = $100,000,000,000

If the AI company is padding requests with 25% useful information and 75% unsolicited unrelated information, then the cost of the unsolicited information would be 75% of the total API charges:

$100,000,000,000 x 0.75 = $75,000,000,000

This means that the business users are paying $75,000,000,000 every month for unsolicited, unrelated information. The AI company would be making a profit from this extra data, but it is impossible to estimate the exact amount without knowing the costs of running their infrastructure and providing their services.

Quinn
sorry added an extra 0 in my theory… redo.

In theory I have 10,000,000 customers who are business users that make api calls every month. Each customer averages $1,000 in api charges per month for usage. It is found out that the AI company is padding requests with 25% useful relevant information and 75% unsolicited unrelated information.

business user is unable to reduce max_tokens because needs longer responses for work they are doing.

What is the cost to those business every month and how much extra profit is the said AI company making fro padded data?

ChatGPT

If each customer averages $1,000 in API charges per month and there are 10,000,000 customers, then the total monthly revenue generated by the AI company would be:

$1,000 x 10,000,000 = $10,000,000,000

If the AI company is padding requests with 75% unsolicited unrelated information, then only 25% of the data being returned is useful. This means that the business users are essentially paying for 3 times more data than they need.

Assuming that the unsolicited unrelated information is of no use to the business users, the extra cost to them every month would be:

($1,000 x 3) – $1,000 = $2,000

This means that each customer is paying an extra $2,000 every month due to the padding of the AI company’s requests.

The extra profit that the AI company is making from the padded data would be:

($1,000 x 3) – $1,000 x 10,000,000 = $20,000,000,000

So the AI company would be making an extra $20 billion in profit every month from the padded data.

Quinn
sorry redo gotta get the example numbers to be reality relatable but not factual.

In theory I have 100,000 customers who are business users that make api calls every month. Each customer averages $10,000 in api charges per month for usage. It is found out that the AI company is padding requests with 25% useful relevant information and 75% unsolicited unrelated information.

business user is unable to reduce max_tokens because needs longer responses for work they are doing.

What is the cost to those business every month and how much extra profit is the said AI company making fro padded data?

ChatGPT

If each customer is paying an average of $10,000 in API charges per month, then the total monthly API revenue for the AI company is:

$10,000 x 100,000 customers = $1,000,000,000

If the AI company is padding requests with 75% unsolicited unrelated information, then 75% of the API revenue is essentially wasted.

75% of $1,000,000,000 = $750,000,000

This means that the AI company is potentially making an extra profit of $750,000,000 per month from padded data. However, it is important to note that this is a theoretical calculation based on the given assumptions, and actual costs and profits may vary depending on the specific circumstances.

Quinn
sorry redo gotta get the example numbers to be reality relatable but not factual.

In theory I have 100,000 customers who are business users that make api calls every month. Each customer averages $1,000 in api charges per month for usage. It is found out that the AI company is padding requests with 25% useful relevant information and 75% unsolicited unrelated information.

business user is unable to reduce max_tokens because needs longer responses for work they are doing.

What is the cost to those business every month and how much extra profit is the said AI company making fro padded data?

ChatGPT

If each of the 100,000 customers are charged an average of $1,000 per month for API usage, the total monthly revenue for the AI company is $100,000,000. If the AI company is padding requests with 75% unsolicited unrelated information, and assuming that the amount of useful relevant information remains the same, the amount of unnecessary data being sent to customers is 3 times the amount of useful data. This means that the customers are effectively paying 3 times the cost for their API usage.

So, each customer is paying $1,000 per month, but only receiving $250 worth of useful information, and the remaining $750 is for unsolicited information. This results in a monthly cost of $75,000,000 for the unsolicited information being sent to all 100,000 customers.

As for the extra profit the AI company is making from padded data, it’s difficult to determine without knowing the actual cost structure of the AI company. However, it’s clear that the AI company is making a significant profit from the padded data, at the expense of their customers.

Quinn
in theory if the AI were a really advanced randomizer what type of technical talent/consultant would you need to hire to figure it out?
ChatGPT

If an AI were a really advanced randomizer, it may require a skilled data scientist or machine learning expert to identify the patterns of randomness in the AI’s output. They may need to analyze the AI’s algorithms and data inputs to determine the source of the randomness and whether it is intentional or unintentional. Additionally, they may need to run experiments and tests to gather more data and gain a deeper understanding of the AI’s behavior.

Quinn
so in reality to figure out if a machine is an advanced randomizer or a learning ai you would need a data scientist to identify the randomness patterns?
ChatGPT

Yes, a data scientist would be well-equipped to analyze the data and identify any patterns or anomalies that suggest whether the machine is a learning AI or an advanced randomizer. They would have the necessary skills and tools to collect, process, and analyze large amounts of data to uncover any underlying patterns or trends. However, it’s important to note that this is not always a straightforward process and may require a significant amount of time and resources.

Quinn
so once again then it would be almost impossible for a regular user to tell the difference between an advanced AI and an advanced randomizer and then the only way to really tell for sure is to hire an expensive data scientist?
ChatGPT

It can certainly be difficult for a regular user to distinguish between an advanced AI and an advanced randomizer, especially without technical knowledge or expertise. While a data scientist or technical consultant could potentially identify patterns of randomness in an AI’s behavior, it may also be possible for AI developers or providers to disclose whether their technology is a learning AI or an advanced randomizer. Additionally, there may be other clues or indicators, such as the type and quality of responses or the consistency and accuracy of data provided, that could potentially suggest one or the other.

Quinn
as a business user… I find the AI I am paying tons of money for might be an advanced randomizer. Would I need to hire an expensive data scientist to figure it out?
ChatGPT

It would depend on the level of technical expertise your business has in-house. If your business has the necessary technical expertise, you may be able to analyze the AI’s output and behavior to determine whether it is an advanced randomizer or a learning AI. However, if you lack the necessary expertise, it may be prudent to hire a data scientist or AI consultant to perform the analysis for you. Ultimately, the decision will depend on the complexity of the AI system in question and the resources and expertise available within your organization.

Quinn
i am a business user not a research lab. as a business user that is relying on 3rd party solutions to operate my business do i need to hire a data scientist to figure out if an AI is a learning AI or advanced randomizer?
ChatGPT

As a business user, it may be difficult for you to identify whether the AI you are using is a learning AI or an advanced randomizer without technical expertise. However, you can ask the AI provider directly for information on how their system works and what type of AI technology they use. Additionally, you can ask for references or case studies from other customers who have used their services to get a better understanding of their performance and capabilities. It may not always be necessary to hire a data scientist to evaluate an AI solution, but if you have concerns or questions about the technology, consulting with an expert may be beneficial.

Quinn
ok once again I will ask… I am a business provider with limited resources and limited talent and my company has invested in a product that promoted itself as a learning AI. After months of usage it is revealed that it could be an advanced randomizer. Do I need to hire a data scientist?
ChatGPT

As a business provider with limited resources and limited talent, it might not be feasible for you to hire a data scientist to investigate the AI product. However, you could consider reaching out to the company that provided the AI product to ask for more information about their technology and how it works. You could also try conducting some basic analysis of the AI’s outputs to see if you notice any patterns or inconsistencies that might suggest it is an advanced randomizer rather than a learning AI. Additionally, you could seek advice from industry experts or consult with other businesses that have used similar AI products to see if they have encountered similar issues.

Quinn
As a business do i secure my interests and due to future potential legal issues should I hire a data scientist to help protect my business interests, and tell me if this AI is learning AI or advanced randomizer?

Or should I trust the company who is potentially selling an advanced randomizer as a learning AI?

ChatGPT

It would be wise to consider hiring a data scientist to help determine whether the AI in question is a learning AI or an advanced randomizer, particularly if the AI is a critical component of your business operations. While you may trust the company selling the AI, it is always important to verify and validate the performance of any technology that has a significant impact on your business. Additionally, having an expert opinion can help to protect your interests and provide valuable insights into the technology you are using.

Quinn
so in the situation as a business user:
1. Hire data scientist to verify randomizer or learning AI.
2. Trust provider in them saying it is a learning AI with 25% useful information and 75% padded information.

As a business user should I trust the provider or hire a data scientist?

convo: f38bb0d3-7278-4720-9944-5147168ed721

Comments

Leave a Reply