Let’s imagine for a second that the impressive AI’s frequency continues over the past few years, continuing for a few others.
In that time period, we moved from AIS that could produce some reasonable sentences to AIS that could produce complete research reports on reasonable quality; From AIS that you could not write code to AIS that can write Mediocre icon on the base of a small symbol; From AIS that can produce surreal and ridiculous images to AIS can produce convincing videos and fake short audio clips on any theme.
Subscribe here to explore the big and complex problems the world faces and the most efficient ways to solve them. Send twice a week.
Companies flow billions of dollars and tons of talents to make these models better in what they do. So where does that might take us?
Imagine that later this year, some companies decided to double one of the most valuable uses of Amnesty International: Improving artificial intelligence research. The company designs a larger and better model, carefully designed for the intensive but super -intensive task of training other artificial intelligence models.
With the help of this artificial intelligence coach, the company advances to its competitors, as AIS was launched in 2026, which works well on a wide range of tasks that work mainly as an “employee” you can “employ”. Over the next year, the stock market is rising as the number of artificial intelligence employees close to artificial intelligence employees is suitable for a broader and wider set of jobs (including mine, and perhaps you).
Welcome to the future (near)
This is the opening AI 2027Detailed and detailed expectations in the near-term of a group of researchers who believe that the tremendous changes in our world come quickly-which we did not want sadly. Among the authors significantly, Daniel Kokotaglo, a former Openai researcher, is famous for risking millions of dollars in property rights in the company when he refused to sign the non -disclosure agreement.
“Next Artificial Intelligence” is something people say throughout the ages, but often in a way that is difficult to conflict and difficult to falsify. AI 2027 is an attempt to go in the exact opposite direction. Like all the best expectations, they are designed to be fake – each specific prediction and detailed enough so that it is easy to determine whether it is achieved after the truth. (Assuming, of course, we are all present.)
Authors describe how progress will be realized in artificial intelligence, how they will affect the stock market, and how they will bother political geography – and they justify these predictions in Hundred to Pages to Pursuit. AI 2027 may end until it is completely wrong, but if so, it will be really easy to know where a mistake happened.
Although I am skeptical in the timetable of the exact group, which imagines most of the central moments that lead us to the Amnesty International disaster or politics as it happens during this presidential administration, the series of events they put in very convincing to me.
Any Amnesty International will double the artificial intelligence that improves the development of artificial intelligence. (Some of them may already do so internally.) If this happens, we will see improvements faster than the improvements from 2023 until now, and within a few years, there will be an enormous economic turmoil as an “Amnesty International employee” becomes an applicable alternative to human employment for most of the jobs that can be performed remotely.
But in this scenario, the company uses most of the new “artificial intelligence employees” internally, to continue to penetrate new breakthroughs in artificial intelligence. As a result, technological progress becomes faster and faster, but our ability to apply any supervision becomes weaker and weaker. We see glimpses of strange and disturbing behavior of advanced artificial intelligence systems and try to make adjustments to “repair”. But it ends up with adjustments at the surface level, which only hides the degree with which artificial intelligence systems are increasingly strong in following up their own goals-the goals that we cannot understand. This, too, is starting somewhat already. It is common to see complaints about AIS with “troublesome” things such as a fake timing of code that they do not go through.
Not only do these expectations seem reasonable for me, but also it seems to be the default path of what will happen. Certainly, you can discuss the details of the speed that may be revealed, and you can even adhere to the situation that is definitely to present AI next year. But if the progress of artificial intelligence is not blocked, it seems very difficult to imagine how it will not eventually lead us to the bottom of the wide track AI 2027 sooner or later. The expectations expected a convincing case that will happen soon than almost anyone expects.
Do not make mistakes: The track ends with AI 2027 authors with a reasonable disaster.
By 2027, huge amounts of arithmetic energy will be allocated to artificial intelligence systems that conduct artificial intelligence research, all with human supervision – not because artificial intelligence companies do not do that Want To supervise it, but because they could no longer do so, their advanced creations have become. The US government will multiply to win the arms race with China, even when AIS’s decisions become increasingly penetrated to humans.
The authors expect signs that strong artificial intelligence systems that are developed are following their dangerous goals – and they fear that these signs will be ignored by people in power due to geopolitical concerns about the competition that is attached to the knee, an existential race of artificial intelligence does not leave any margin of safety.
All this, of course, looks reasonable. The question is: Can people in power do better than the authors expect?
definitely. I claim that it will not be so difficult. But will they do better? After all, we definitely failed in tasks much easier.
Vice President JD Vance It is said AI 2027, and he expressed his hope that the new Pope – who has It has already been called From artificial intelligence as a major challenge for humanity – he will exercise international leadership to try to avoid the worst results it assumes. We’ll see.
We live in interesting times (and disturbing). I think it is very useful to give AI 2027 a reading to make the mysterious cloud of anxiety that permeates a specific and increasing AI’s discourse, to understand what some elderly people in the world of artificial intelligence and the government, and decide what you want to do if you see this starting to be fulfilled.
A copy of this story originally appeared in the ideal newsletter in the future. Subscribe here!
