Influencer marketing trends for 2025
Posted: Tue Jan 21, 2025 6:39 am
What are the limits of AI?
There are two aspects. When I train my colleagues, their first reaction is that it is magic and can do anything, including taking their jobs. But after using it, they think that, in the end, it is horrible because it spells badly and needs to be guided. They understand that AI is not automatic or magic, but needs to be informed and trained. Once trained, AI becomes powerful. The limit is on us; if we do not train it well, it will not correct our limits.
However, machines can work 24/7, automatically moderating social media, which is magical. There are physical limitations, though; they won't help you drive a car. However, when applied intelligently, AI can make redundant and painful tasks less painful.
AI can revolutionize research. Instead of manually reviewing numerous articles to gather information, you can ask AI to quickly synchronize and analyze numerous sources. This was not possible before.
Social ?media
It’s already there. Facebook’s algorithm decides what content appears in your news feed based on what it knows about you. It’s not necessarily better, but it works based on your interactions. When I first started using Facebook, you saw 100% of your friends’ posts — it was great. Today, algorithms prioritize content based on engagement and revenue. An ideal social network would allow users to reset and personalize their feed according to their current interests, filtering out unwanted content. That’s what TikTok is trying to do — but not perfectly. Hyper-personalization is good, but users need to be able to control the content they see.
On the contrary, what we see today on social media is that when you post a message, we suggest that you use AI to write the message, generate the visuals, create videos, titles, etc. This is where I see a huge bias because these AIs are trained from the content of social platforms and if the person is not careful, they will reproduce what has already been done. We will find ourselves in a kind of algorithmic bubble where it will be very difficult to stand out. What we love about social media is discovering accounts with a personal touch that makes us want to come back and subscribe. Once you have decided on your editorial line on LinkedIn or Facebook (whether it is humorous or otherwise), if it is automatic, it will no longer be the same.
What worries me most is that the content generated may be conditioned by a context and rules managed by social platforms. If I want to publish content that the platform may consider offensive, I will not be able to do so. But there is a big advantage: at least there will be no content with spelling and grammatical errors. I see a huge advantage in the use of AI in social media. It will prevent us from seeing content that hurts our eyes!
Stay ahead of the curve in influencer marketing by downloading our free study now, packed with up-to-date data, creator rankings, interviews, and key metrics. Let “Influencer Marketing Trends for 2025” be your essential guide to stay ahead of changes in the industry.
Download now
Many people are talking about the idea of synthetic social media, where we interact with robots and AI-generated content as much as with real people. What do you think?
It's complex. If the content is entertaining, it doesn't matter where it comes from. But the probl vp media email database em arises when fake content is presented as real. Just like in reality shows, where reactions are staged, it's misleading. I don't mind, but for young audiences it's crucial to distinguish fake content from real. AI-generated educational content needs to be accurate to avoid misinformation.
I am more concerned about artificial general intelligence (AGI), which can learn and adapt independently, which poses significant risks. Broadly speaking, an AGI is a multi-AI system that can understand, make decisions, control solutions, and continue to learn, without being controlled. On this point, I prefer to side with the pessimists, because the risk is enormous.
Electoral manipulation is not so much a question of content production as of whether platforms have a responsibility to distribute this content.
Scientists have been talking about AGI for over 50 years, how close is it to becoming a reality?
ChatGPT's progress suggests that AGI could be shut down, possibly within two years, but we need assurances before making it public. The question is which actor will be able to develop an AGI capable of taking advantage of the accessible data and implementing protective measures.
When OpenAI introduced SORA, its tool for generating hyperrealistic videos, they said: “That’s great, we have the most advanced solution on the market.” Shortly afterwards, they announced that they would not be giving access to the general public for the time being, because they did not know what people could do with it and for what purposes they could misuse it. Cynicism or protection measures in case of error?
There are two aspects. When I train my colleagues, their first reaction is that it is magic and can do anything, including taking their jobs. But after using it, they think that, in the end, it is horrible because it spells badly and needs to be guided. They understand that AI is not automatic or magic, but needs to be informed and trained. Once trained, AI becomes powerful. The limit is on us; if we do not train it well, it will not correct our limits.
However, machines can work 24/7, automatically moderating social media, which is magical. There are physical limitations, though; they won't help you drive a car. However, when applied intelligently, AI can make redundant and painful tasks less painful.
AI can revolutionize research. Instead of manually reviewing numerous articles to gather information, you can ask AI to quickly synchronize and analyze numerous sources. This was not possible before.
Social ?media
It’s already there. Facebook’s algorithm decides what content appears in your news feed based on what it knows about you. It’s not necessarily better, but it works based on your interactions. When I first started using Facebook, you saw 100% of your friends’ posts — it was great. Today, algorithms prioritize content based on engagement and revenue. An ideal social network would allow users to reset and personalize their feed according to their current interests, filtering out unwanted content. That’s what TikTok is trying to do — but not perfectly. Hyper-personalization is good, but users need to be able to control the content they see.
On the contrary, what we see today on social media is that when you post a message, we suggest that you use AI to write the message, generate the visuals, create videos, titles, etc. This is where I see a huge bias because these AIs are trained from the content of social platforms and if the person is not careful, they will reproduce what has already been done. We will find ourselves in a kind of algorithmic bubble where it will be very difficult to stand out. What we love about social media is discovering accounts with a personal touch that makes us want to come back and subscribe. Once you have decided on your editorial line on LinkedIn or Facebook (whether it is humorous or otherwise), if it is automatic, it will no longer be the same.
What worries me most is that the content generated may be conditioned by a context and rules managed by social platforms. If I want to publish content that the platform may consider offensive, I will not be able to do so. But there is a big advantage: at least there will be no content with spelling and grammatical errors. I see a huge advantage in the use of AI in social media. It will prevent us from seeing content that hurts our eyes!
Stay ahead of the curve in influencer marketing by downloading our free study now, packed with up-to-date data, creator rankings, interviews, and key metrics. Let “Influencer Marketing Trends for 2025” be your essential guide to stay ahead of changes in the industry.
Download now
Many people are talking about the idea of synthetic social media, where we interact with robots and AI-generated content as much as with real people. What do you think?
It's complex. If the content is entertaining, it doesn't matter where it comes from. But the probl vp media email database em arises when fake content is presented as real. Just like in reality shows, where reactions are staged, it's misleading. I don't mind, but for young audiences it's crucial to distinguish fake content from real. AI-generated educational content needs to be accurate to avoid misinformation.
I am more concerned about artificial general intelligence (AGI), which can learn and adapt independently, which poses significant risks. Broadly speaking, an AGI is a multi-AI system that can understand, make decisions, control solutions, and continue to learn, without being controlled. On this point, I prefer to side with the pessimists, because the risk is enormous.
Electoral manipulation is not so much a question of content production as of whether platforms have a responsibility to distribute this content.
Scientists have been talking about AGI for over 50 years, how close is it to becoming a reality?
ChatGPT's progress suggests that AGI could be shut down, possibly within two years, but we need assurances before making it public. The question is which actor will be able to develop an AGI capable of taking advantage of the accessible data and implementing protective measures.
When OpenAI introduced SORA, its tool for generating hyperrealistic videos, they said: “That’s great, we have the most advanced solution on the market.” Shortly afterwards, they announced that they would not be giving access to the general public for the time being, because they did not know what people could do with it and for what purposes they could misuse it. Cynicism or protection measures in case of error?