ChatGpt is now accessible via WhatsApp. Users can interact directly with the AI of OpenAI through the popular messaging app of Meta.
Adding +1 800-242-8478 to your contacts, you can start a conversation with ChatGpt without the need for registration or additional applications.
This integration is only capable of answering text questions and lacks features such as real-time Internet search and – at least in Europe and therefore in Italy – vocal interaction.
IA
ChatGpt for free on WhatsApp: just register a number in the address book
by Pier Luigi Pisa
December 19, 2024
On WhatsApp, AI knowledge is only updated until January 2022so its ability to provide information on recent events is limited.
But there are other problems, and potential risksto be taken into account if you decide to use ChatGpt on WhatsApp.
Before registering the number released by OpenAI, you should ask yourself the following questions: a more conscious use of artificial intelligence.
Why does OpenAI offer ChatGpt on WhatsApp for free?
The official version is the one provided by Kevin WeilChief Product Officer at OpenAI: “Our mission is to make AI a benefit for all humanity. Bringing ChatGpt to phones and WhatsApp is a step towards this goal.”
The intent is therefore clear: to broaden the pool of users who use ChatGpt as much as possible, also embracing all those who are currently hindered by a lack of confidence with technology or from less powerful devices, such as low-end smartphones.
But what does OpenAI really gain from it?
Most likely, and give.
WhatsApp can count on approx 3 billion active unique users. Approximately more than 140 billion messages are exchanged on the Meta app every day 1.6 million per second.
All these messages contain very precious texts for companies like OpenAI that develop artificial intelligence, since they constitute “quality” data necessary for the improvement and training of AI models of the present and future.
Why is data produced by humans so important to OpenAI?
Without the data produced by humans, ChatGpt would never have been born. This data is extremely valuable to a company developing artificial intelligence because it represents concrete examples of how people communicate, make decisions, solve problems, or interpret information. By analyzing them, AI can learn patterns, contexts and behaviors, thus becoming more precise, natural and effective in its responses or tasks, arriving at imitate the creativity of human beings.
What is meant by “training”/“training” of an artificial intelligence?
It is the process by which a system is exposed to large amounts of data to learn how to carry out specific tasks.
During this process, the AI analyzes the data provided, identifies patterns or rules, and adjusts its internal parameters to improve its performance. For example, a model can be trained with thousands of images to recognize objects, or with texts to understand and generate language.
The goal is for AI to become increasingly precise and effective in carrying out the task for which it was designed.
How is user data used for AI training?
“We retain certain data from user interactions – OpenAI says – but we take steps to reduce the amount of personal information in our training datasets before they are used to improve and train our models. This data helps us better understand user needs and preferences, allowing our model to become more efficient over time.”
IA
100 things to know about artificial intelligence
by Pier Luigi Pisa
June 15, 2024
What does Meta gain from ChatGpt on WhatsApp?
The crittografia end-to-end of WhatsApp ensures that messages are protected during their transit between the user’s device and WhatsApp servers.
However, once the message reaches ChatGpt, the content is decrypted to allow the request to be processed. This implies two fundamental aspects: on the one hand, data transmission occurs securely; on the other, the content becomes accessible exclusively to OpenAI for processing purposes and, potentially, model improvement.
Therefore, from the point of view of training its artificial intelligence, called Meta AIthe company led by Mark Zuckerberg He doesn’t seem to get any benefit from it.
Meta’s interests, in reality, in this case they appear to be cheap.
Last June, the Menlo Park giant announced its intention to integrate free AI-based chatbots on WhatsApp, with the aim of more effectively monetizing its instant messaging app.
Just last June the Wall Street Journal wrote that Meta was planning to “allow businesses to use AI chatbots on WhatsApp.” Exactly what OpenAI has done in recent days using a business account of the instant messaging app.
“Chatbots – explains the Wall Street Journal – will help companies manage the influx of questions they receive from customers. These customer relations chatbots represent an early test of how Meta is looking to integrate artificial intelligence into its applications in ways that could, in the future, generate revenue.”
Does OpenAI preserve conversations with ChatGpt on WhatsApp?
Yes, it’s possible. When you register the ChatGpt number in the address book, and open the chatbot account on WhatsApp for the first time to start a conversation, it appears a warning that should not be underestimated: “You are chatting with ChatGpt, an AI assistant. By continuing, you accept our terms of use and confirm that you have read our privacy policy on openai.com. Conversations may be monitored for security reasons.”
So does OpenAI read the messages that are sent to ChatGpt via WhatsApp?
Yes, it could. On the page dedicated to the FAQ – the most frequently asked questions – regarding ChatGpt on WhatsApp, we read: “We retain and may view WhatsApp messages sent to ChatGpt for a limited period of time for security and abuse prevention purposes”.
Does ChatGpt “store” the information that users provide through WhatsApp?
No. The “memory” function of ChatGpt, which allows AI to collect information about the user based on past conversations, is not active on WhatsApp.
Can OpenAI use WhatsApp messages to train its future AI models?
Theoretically yes. OpenAI says it could use “data submitted to ChatGpt and other services” to improve the performance of AI models.
“For example – specifies OpenAI – based on user settings, we may use user prompts, model responses and other content such as images and files to improve model performance”.
Can OpenAI be prevented from using messages sent to WhatsApp to train its AI?
Yes, but it’s more difficult than what happens on the ChatGpt app or on its desktop version.
In these two cases, simply click on the settings, then on “Data control” and finally on “Improve the model for everyone”. By deactivating this item, you do not consent to the use of your content to train OpenAI’s artificial intelligence models.
Tutorial
ChatGpt, what are the new temporary chats for
by Pier Luigi Pisa
03 Maggio 2024
They are WhatsApp this operation is not possiblesince the settings obviously only concern the use of the Meta app and not individual conversations.
What you can do is make a specific request on the OpenAI portal dedicated to privacy management, or write an email to the address [email protected] in which the data provided to ChatGpt on WhatsApp is requested to be excluded from AI training.
