AI (artificial intelligence) is found in various forms:
- Music streaming like Spotify, Apple Music, Google Music.
- Netlfix, Amazon Prime video.
- Google, Bing search engines.
These are examples of more famous forms of AI. Algorithms analyse large amounts of data where a specific target is presented.Statistical analysis attempts to make a selection that has the greatest probabilities to achieve that goal. Let us take Spotify, they analyse the music you are listening to and try to make a prediction which other artists and music you will love. Each week you get a list of songs that will compose Spotify for you. Same for video streaming services. Google and Bing are search engines that live ads. They optimize search results so that their customers, the advertisers, are happy with the extra sales of services or products. Alone, advertisers will continue to pay for advertisements. Over The last decade, you’ll see the quality of the search results deteriorating. Where information was previously displayed, advertisements are now displayed and then search results from websites that do not advertise but do sell products or services. The percentage of useful search results that have nothing to do with sales is steadily decreasing.
There are also other examples:
- Computer games where opponents have to be screwed up or where you have to race against computer-controlled opponents, who have algorithms that offer a certain level of opposition.
That’s why you choose easy or difficult mode in a game. AI can also make the opponent’s behavior more natural. The analysis can be done in datacenters (not real time) and the AI model (result of the datacenter calculations) can then be performed locally on your device.
The chip, Myriad II (new version Myriad X) has video processing onboard and also a CNN, a convolution neural network. Video processing is closely related to AI by requiring both floating point calculations. Floating point is an English word, floating comma is actually a hard to use translation in the Dutch.
Raytracing, in thet Dutch presumably difficult to translate, was impossible in the past because the calculations needed to calculate the refraction of light and reflection on objects while one plays for example a game were far too bulky. So far, mainly rasterization was used. NVIDIA promotes Turing as ray tracing but the truth is slightly more nuanced. Turing is a hybrid combination of ray tracing and rasterization at lower resolution and limiting the number of light sources (to limit calculations) after which tensor processors (neural network) lower resolution to a higher resolution Translates based on an AI model. GPU芒 鈧劉 s have brought AI to this level because AI and GPU芒 鈧劉 s FP16 and use FP32 calculations. But AI evolves to lower precision (FP16, FP8) while GPU evolves to higher percision (FP32, FP64).
AI is already in use in many places (mostly to optimise sales) but will grow to interest.Hardware (chips) will be the way to perform an AI algorithm in the fastest and most efficient way (number of watts per FLOPS or TOPS). ASIC (Application specific Integrated circuit = a chip specifically created for a particular application) is the key to AI. The struggle in this goes between China and the United States. Europe is not playing with it, only as a tool for both superpowers. A key strategic flaw in Europe. If we look at what the hated bottle represents Europe, we should not look far to explain the lack of understanding and strategy. If something happens in Europe (individuals, not the government), Movidius (Myriad) or DeepMind then this prompt will be sold to the Americans.
The consequences of this strategy will be demonstrated within ten to twenty years.
Thanks, Quora user, to ask my answer on this interesting question!
By using Quora you make use of AI/Machine Learning without having to go through it.
The next post (in English) is very interesting if you want to know more about how Quora puts this technology in:
The answer is in what AI really is now. AI stands for Artificial Intelligence, artificial intelligence, intuitively feels like something human, as it can reason.However, what we see, experiencing, as modern AI is just predictive.
The current practical definition of AI is to be able to predict behavior by analyzing (very much) behavior in the past.It’s just a statistical game, no more.
Given a lot of information about the past, we can make quite reasonable predictions about the future.However, this is nothing new, especially for advertising purpose this has been happening for a long time. Every time you see an advertisement, be it a booklet, a poster, an email, you get to see something influenced by AI, which uses AI.
Using a spell checker, this often makes use of the knowledge that certain words and certain types of errors are more common than others.If you are sitting at a bank, statistical analyses are used to determine your interest.
You will encounter it everywhere in all different forms.You can’t actually use a computer without a way to influence AI. Without using AI directly or indirectly.
Recognizing faces, images, reading text from a photo, converting speech to text, translating text so you can randomly read which internet page in your own language, offering advertising, chess, playing poker, investing, Speed controls.
You might also be able to buffer text if you predict what the user is going to do, I do not think it is.
AI is already used by many large companies like Google, Uber, Facebook to automate certain processes.Take for example the advertisements you see on the Internet, these are chosen for you based on a lot of data about your surfing habits. They do this with artificial calculation models (AI).
Another example is the videos you get recommended on YouTube.By analyzing data about your viewing habits, Google (owner of YouTube) can estimate which other videos may be of interest to you.
A final example, the Daily playlists in your Spotify.These are also automatically generated for you based on AI models.
Secretly, a lot of AI models are running to send people on the Internet.