Artificial intelligence has been moving the world more and more in recent years. AppHowever, it still hasn't managed to jump on this fast-moving train in a way that would allow it to be considered a full-fledged AI passenger, let alone a trendsetter. Its AI functions are still quite weak compared to its competitors, for which it has been receiving harsh criticism from all sides. No one really understands how a company of its caliber could afford such a mistake. The Information portal has now tried to provide some answers to these questions, having managed to do so by interviewing a number of anonymous current or former employees.anců AppI can find out a lot of information about the development of AI.
You could be interested in
In development Apple Intelligence he works Apple for a very long time, initially considering whether it would not be best for its operation to create both a small and a large language model from which its "smartness" would draw, while the small one would run directly on iPhonech, big then na cloudManagement Applu, however, later leaned towards creating one large and as complex a language model as possible, which would run on cloudbut it can handle everything that is required of it. Unfortunately, the first problem in the development has already begun to take root here, as it was not possible to specify for a long time what the given model should include, how complex it should be, and so on. It was precisely because of indecision or sudden changes in plans that some top employees hadanci AppI want to leave.
However, these were not just problems with development as such. Many employeesanců Applu mentioned that during development, the company had a very bad working atmosphere stemming from poor management, which allowed for an overly relaxed corporate culture, combined with zero ambition from managers or an absence of risk appetite regarding features that were complex but could be revolutionary. The combination of these factors slowed down development considerably, with one development team running into a problem and passing the project on to another team to solve the problem. Siri and other AI functions were being passed around from team to team like a hot potato, which stalled development. Then came the classic employee issues like dissatisfaction with salary, low vacation time, and so on.
Gallery
ChatGPT as an (un)interesting opponent
A major turning point in development could have been 2022, when the company debuted OpenAI with your weakness ChatGPT. Apple However, he brushed this project off the table, saying that the boss Apple AI John Giannadrea claimed that chatboots type ChatGPTs do not have much added value for users and therefore will not see a boom. On the contrary, the solution AppThe software that was being worked on at the time was supposed to be exactly what people wanted, while also being relatively easy to create. "Stupid" Siri According to Giannandreay's vision, the improvement would have been "just" the deployment of better training data, which would have allowed it to catch up with the competition. But the result was not achieved.
Although ChatGPT has been growing stronger practically from the very beginning, Apple deliberately ignored it and in 2023 it came from the management Appa regulation that engineers may not include any model from companies other than Applu. It was evidently still believed that it was Apple able to develop their own solution that will minifully competitive for ChatGPT and similar models. Even then, developers AppThey knew that the model they had created was not working nearly as well as the solution OpenAI in direct comparisons.
Small changes last a long time...or not.
About improvement Siri people in AppThey tried all the time on multiple fronts, while miniAt least one of them was doing quite well. The current boss's group Siri Robby Walker, for example, was able to achieve his goal through the transformation of an activation phrase Siri from "Hey Siri" on "Siri", which took her two long years. Unfortunately, other ongoing projects that this group was working on were brushed off the table by the management. For example, it was expected that she would be able to Siri analyze the user's emotions and communicate with them based on that. In other words, she was preparing a kind of psychologically more advanced Siri able to help in an emergency as needed.
It was also expected that they would go through Siri to control in Vision Pro applications, perform various tasks in it, browse websites including changing font size, windows, etc. However, this did not fall on fertile ground either, because Siri She simply wasn't able to achieve similar interactions, and most importantly, it wasn't clear whether it would even be possible.
You could be interested in
WWDC 2024 like a blow to the waist
You remember the WWDC 2024 and how Apple spectacularly showed off his skills Apple Intelligence a Siri for example, in the form of her reading emails, thanks to which she was subsequently able to find flight data in real time, provide lunch reminders, offer navigation to a given location via Apple Maps and so on? So forget about such use. Virtually the entire demo was allegedly completely fictitious and its release was supposed to be a big surprise for all the developers involved in the product. They had never seen these features in working condition and probably didn't even believe they would.
The funny thing is that according to sources, the only function from this complex demonstration of the ability is actually Siri interact with the system a pulsating color strip around the perimeter of the display. That was the only thing that was supposed to work without any problems and was reliable. It was this deceptive presentation that was a great disappointment for many developers, because it showed that Apple significantly deviated from its presentation policy in the form of a demonstration of functions that are miniare at an advanced stage of development and testing, and therefore their launch is likely nestthere is absolutely nothing in the way. It is therefore evident that Apple from charger failure AirPower did not teach.
Flash forward to better times?
At the moment, according to sources from the aforementioned portal, the mood is Applu is already a little better, as the new leadership of AI projects, led by the legendary Craig Federighi and Mike Rockwell, is to add to the projectoneFederighi should have taken the lead in solving the problem and instructed the developers to do absolutely whatever it took to get it done. Apple products to run the best AI features imaginable. In other words, open source language models are green and who knows what else. Whether this new strategy will bear fruit in the short and long term, of course, only time will tell.
But again we have lots of important emojis🤣🤣🤣