Robotic intelligence. AI in the shadow of Web3, AI jokes Artificial intelligence (AI) is entering life and the labor market at a tremendous pace, but it should be noted that it has both many advantages and potential threats.
Many are convinced that AI is a remedy for the digital divide. But is this not an overly naive conclusion?
Robotic intelligence, let’s try to find out.
Bill Gates, who participated in the World Economic Forum in Davos, said in an interview with CNN that the development of artificial intelligence in the next five years will be no less comprehensive and stunning in its consequences than the machine revolution in agriculture at the beginning of the 20th century.
Thanks to this, for the first time in the world (and not only in the most developed countries), millions of former farmers began to move to cities, learning previously unimaginable professions.
AI, Bill assured, will bring tangible benefits to humanity in this relatively short period of time, as it will be able to provide highly personalized and very practical advice in many areas.
From writing program code and learning new professions to improving personal health.
At the same time, according to Bill, access to such beneficial services in terms of increasing the general standard of living will be almost equally available to residents of all countries where, in principle, the Internet is available.
If anything, consultations with artificial intelligence do not require special equipment, except for the most common computer or smartphone. A very positive assessment, but I did not expect anything else from Bill.
The decline or decline of social networks.
Andrew Golis, chief content officer of New York Public Radio, described in an emotional note to the magazine “The Nieman Journalism Lab” how AI-generated content will spread across social media and traditional digital media in the near future.
As a result, people who want to maintain at least some objective view of what is happening will return to the times of extremely limited web communities, the content of which will be created independently.
Or rather, sponsored by members of these communities or journalists of closed communities.
At the same time, the main reason for the rapid cooling of the masses towards social networks, according to the expert, is not so much the multiplying deep fakes, but the AI content optimization algorithms.
If you did not know, then AI feeds the participants of these systems the most suitable (based on their posts, likes, web surfing history, etc.) messages to the analysis of a neural network.
As a result, the news feed, which fully corresponds to the interests and expectations of a person, gradually turns into a soothing, but completely uninteresting white noise for him.
Therefore, it is much easier to give up regular viewing of news on social networks than the owners of social networks would like.
Most of your social friends are AI created with a specific idea: to direct your thoughts in the interests of some group. I don’t think that correspondence, or even a conversation (for example, WhatsApp) with AI can be enjoyable.
“Edition Inc” provides dry statistics. According to Crunchbase, in 2023, Web3 startups worldwide attracted $7.3 billion (while in 2022 – $21.5 billion). However, global investments in the artificial intelligence industry, according to Dealroom estimates, reached $17.8 billion.
Experts point out that Web3, with all its theoretical advantages, has not been able to offer a single truly attractive application to mass users. Especially considering the fatigue that many still feel after the “Covid” quarantines from purely digital communication (the so-called “Zoom fatigue”).
At the same time, modern, widely available AI with its chatbots, image generators, audio and video often fascinates even those who truly criticize it.
Influencer sounds cool.
Social networks, the number of whose posts, according to “Morning Consult”, is already steadily decreasing, have suffered a new disaster.
As I already wrote, what is caused by generative AI? An article from the “Financial Times” states that large brands that are traditionally active on social networks (for example, “Olaplex” and “Victoria’s Secret” were mentioned).
Large brands are increasingly ordering advertising content not from real people, but from virtual AI advertising “people”, whose accounts are confidently gaining popularity.
The turnover of such content in global social networks over the past year is estimated at 21 billion dollars, and at a rate of about 1 thousand dollars per marketing message.
The AI character clearly wins over biological competitors both in its readiness to provide the requested content at any time of the day, and in the full compliance of the posts created by it with the advertiser’s requirements.
Thus, an analysis of the activity of virtual influencer XXX in promoting the H&M brand showed that his messages were viewed by 11 times more users than traditional advertising.
All this despite the fact that (or maybe just because?) many social media users call AI-generated opinion leaders “fake” or “manekenies”.
I think you are starting to understand what robotic intelligence is, it is the fusion of human thought with AI and the decisive factor here is artificial intelligence.
Convincing multimodality.
It is no secret that Bill Gates interviewed none other than OpenAI CEO Sam Altman, who was removed from his post by a decision of the board of directors. What is interesting is that just a few days later, miraculously, and with great online help from millions of ChatGPT enthusiasts, it returned.
The conversation was devoted to the future of artificial intelligence, which, according to Altman, will be determined by multimodal models in the next few years. They can equally easily cope with speech recognition and voice synthesis, text blocks, static images, music and video.
The main thing that developers of generative models need to achieve in this interval is reliable, highly reproducible reliability:
“If you ask GPT-4 the same question many times, one of its answers will probably be really good. The AI system itself cannot yet evaluate this, only the one who asks it evaluates. We need to ensure that the generative model always answers the question asked once in the best way from the point of view of human perception.
Sam Altman also said that the artificial intelligence revolution will undoubtedly be “the fastest of all technological advances in the entire previous history of mankind”.
By the way, it quietly mentioned that society will have to quickly adapt to the pace of further changes, including due to rapid development. In his opinion, robotics will be, as it were, a link between the machine mind and the material world.
Without unnecessary emotions and speculation, it is clear that humanity will stop thinking and drawing logical conclusions about what is happening around. The vast majority will rely on what AI says! Robotic intelligence, in place of the finale.
I will say right away that the article was created without AI intervention.
The question may arise, how to distinguish a human-written text from an AI-“invented” text.
I have noticed that one of the most serious problems that prevents artificial intelligence robots from fully conducting “truly human” dialogues with living people is the extremely low level of understanding.
Did anyone ask?
Quite simply, with large models of humor, sarcasm, metaphors and other indirect ways of expressing thoughts. If anything, which carriers of biological intelligence use to a greater or lesser extent and confidently master at a fairly early age.
AI does not yet fully know how to joke and spit out clear and understandable pearls of wit.
WAIT! However, it can!
I asked AI a question directly:
Do you know how to joke, show it with an example?
Joke: Why does a computer never go on dates?
Answer: Because it’s afraid of love viruses!
Joke: Why does a tomato blush?
Answer: Because it saw ketchup undress!
Joke: What did a haystack say when someone touched it?
Answer: “Go away, you’ll crush me!”
Joke: Why does a robot never forget birthdays?
Answer: Because it has great memory – RAM!
Joke: Why do programmers never go outdoors?
Answer: Because there are too many bugs! I hope this cheered you up!
Want to hear more?
How do you work, how much processor do you use and how much memory do you need? I work as a cloud service and do not physically work on one specific computer.
However, if you are interested in how I work, I can give you a rough description:
1. Processor: My models are run on specialized server platforms that have high-performance graphics processing units (GPUs), such as NVIDIA A100 or similar cards. GPUs are optimized for machine learning tasks.
2. Processor usage: When I answer questions or perform tasks, only as much processor resources are used as are needed at that particular moment. This could vary greatly depending on the complexity of the task.
3. Memory: My models, such as GPT-4, are trained using a large number of parameters (hundreds of billions), and these parameters are stored on servers.
The model itself requires a significant amount of RAM to run the model components. Typically, this is tens to hundreds of gigabytes of memory.
Hello, robotic intelligence!
