Advanced Generative AI: Back to The future With Apple's Knowledge Navigator?
Apple's 1987 concept video illustrates the future of advanced generative AI.
Although I primarily write articles about autonomous vehicles, my research into artificial intelligence leads to other insights. Recently, I explored the next steps to advanced generative AI, returning me to one of my favorite videos: The 1987 Apple concept video called The Knowledge Navigator.
There are three advanced concepts for generative AI demonstrated in the video:
Integrated Generative AI. Integrated generative AI integrates on-device (iPhone, iPad, Mac, Apple Vision Pro, or potentially new devices) data/applications with cloud-based and internet-based AI.
Edge AI. Edge AI is an emerging opportunity to incorporate selected AI functions into devices, providing significant advantages.
Interactive Voice Communications. Finally, generative AI will use interactive (back-and-forth) voice communications, going beyond the one-way voice communications we employ today.
Collectively, these advanced AI capabilities will eventually change the way we use computers, making what we do today seem primitive.
Interestingly, Apple previewed these capabilities in 1987 with the Knowledge Navigator concept video, and watching this video is the best way to visualize what is coming. This concept video presented a futuristic view of artificial intelligence. Central to this vision was a device resembling a modern iPad but with significantly enhanced features. It included a touchscreen interface and an AI assistant represented by a human-like avatar. This assistant was designed to comprehend and reply to spoken language naturally, making the device highly interactive. The video showcased the Knowledge Navigator as an academic tool for research, data analysis, managing schedules, and facilitating communication.
Here is a link to the original video, which you need to watch to appreciate it: Knowledge Navigator 1987
The video demonstrated several forward-looking technologies and capabilities. The most prominent was the voice-controlled interface, allowing users to interact with the device using spoken commands. This AI assistant could carry out tasks, engage in conversations, and display emotions. Modern voice-activated assistants like Siri and Alexa are primitive versions of this. Another notable feature was its ability to access, integrate, and analyze vast amounts of data, pulling information from various sources, similar to generative AI. Many of these capabilities predicted in the 1987 video are available today, but what's most interesting is the way it applied integrated generative AI and interactive communications.
The Knowledge Navigator concept video from Apple in 1987 didn't explicitly predict the development of large language models of AI as we understand them today. However, the video did envision advanced features in artificial intelligence that resonate with the capabilities of generative AI, including aspects that are foundational to large language models. Most importantly, it did this by integrating what was, most likely, on-device and cloud data using an Apple device.
The Knowledge Navigator showcased an AI assistant with three key features:
Natural Language Processing (NLP): The assistant understood and responded to natural language, much like today's large language models. Although the video didn't explain the technical details, it highlighted the importance of NLP for AI's ability to interact in a human-like manner.
Voice Recognition and Interaction: The AI could recognize and respond to spoken words, similar to how current large language models can process and produce human-like text and speech. More importantly, it demonstrated interactive voice communications.
Data Retrieval and Integration: It showed the capability to access and combine information from different sources, resembling how modern AI large language models use extensive training data to generate informed responses.
While the Knowledge Navigator didn't accurately predict the scale or complexity of today's AI models, it did foresee essential elements of AI interaction and language understanding. Its emphasis was on AI's user experience and practical applications rather than the specific technologies that would later become fundamental in AI development.
The video elicited diverse reactions, mirroring the varied perspectives of that time on the future of technology. Many saw it as a visionary insight into the future of computing and technology. They appreciated it as an inspiring portrayal of potential achievements in human-computer interaction, artificial intelligence, and personal computing. The depicted digital assistant was viewed as a forerunner to subsequent innovations like Apple's Siri, Amazon's Alexa, and Google Assistant.
Conversely, some critics considered the video excessively optimistic and not grounded in reality. They argued that the technology shown was too advanced for the era, creating unrealistic expectations about the speed and direction of technological progress. Given the limited AI and computing capabilities of the late 1980s, these critics dismissed the video as mere science fiction.
Now, let's explore these three advanced generative AI capabilities, and we can also see Apple's competitive advantages.
Integrated Generative AI
Today's generative AI applications primarily leverage the vast amount of information available on the internet to process user queries and respond. Additionally, these applications can be utilized to process attached documents, which is particularly helpful for tasks like editing and summarizing.
Integrated generative AI represents a more advanced step, as it can access information seamlessly from various sources, including local devices, applications, remote servers, the cloud, and the internet. This comprehensive access will enable much more powerful capabilities.
The best way to describe how integrated generative AI would work is through some examples.
Integrated Generative AI Examples
Here are some examples of integrated generative AI illustrating how a response to a request can use local device-specific data, the cloud, and the internet.
Hey Siri, find the research paper I wrote 18 months ago on the potential impact of the new weight loss drugs. Then, could you update it with the latest research and drug trial results, add quotes from food industry executives, and include any of the latest projections?
Hey Siri, create an analysis of my investment portfolio by adding all the buy/sell ratings and target stock prices from the leading analysts to each of my stocks.
Hey Siri, I have a meeting with the members of my active research group coming up. Can you find and list all references to new publications by each member so I can mention them to them?
Hey Siri, someone asked me for a list of my favorite restaurants. Can you review my calendar for the last two years, summarize each restaurant, include how many times I went there, and add a couple of ratings for each one?
Hey Siri, here are the results of my current blood tests. Please prepare a trend comparing results from my last four tests, add expert opinions for any results outside of normal ranges, and prepare a list of questions I should review with my doctor.
Hey Siri, please categorize all my spending for last year and compare it to what financial experts typically recommend.
Hey Siri, my friends the Carters and Smiths asked me for movie recommendations. Access the app I use to rate movies and draft emails to them with my ratings. Also, include a summary of each movie and compare my ratings to the most popular ratings.
Hey Siri, I'm interested in the new AI product management position for AI at Apple. Rewrite my latest resume to emphasize that experience related to that specific position and draft a cover letter incorporating some of the latest information on Apple's AI efforts.
Apple Integrated Generative AI Advantages
Apple has several significant competitive advantages in integrated generative AI.
The largest device ecosystem. Apple has more than two billion active users of its devices (Apple Reports 2 Billion Active Users). This gives it a considerable advantage in access to the device end of integrated generative AI.
Software integration: Apple's tight integration between hardware and software, evident in platforms like iOS, iPadOS, macOS, and watchOS, presents a potential advantage for seamlessly integrating generative AI features into user experiences across their diverse product range.
Hardware capabilities: Apple is renowned for its hardware design and manufacturing. Powerful, energy-efficient chips like the M1 and M2 series CPUs and Neural Engine NPU in their devices provide a strong foundation for running complex generative AI models efficiently, and it will most likely develop AI-specific processing chips.
Privacy focus: Apple emphasizes user privacy in its products and services. This could give it an edge in developing and deploying generative AI applications prioritizing user data security and control, potentially appealing to privacy-conscious consumers.
Developer ecosystem. Apple has a vast ecosystem of developers creating applications for its devices. They can integrate generative AI into their applications and accelerate introduction. Apple's machine learning research team released MLX, a machine learning framework where developers can build efficient models on Apple Silicon and deep learning model library MLX Data.
Edge AI
The next important advanced generative AI capability is AI at the edge, also known as edge AI. This refers to deploying artificial intelligence algorithms, AI preprocessing capabilities, and selected AI models directly on devices rather than processing data on centralized cloud servers. Edge AI provides several advantages:
Faster processing: Processing data locally on a device reduces latency and allows quicker responses, especially in time-sensitive applications. Processing data locally can save on bandwidth and cloud computing costs.
Improved privacy: Private data on a device doesn't need to be sent to the cloud or could be encrypted before sending requests to the cloud. Privacy concerns and data security risks are significant, and edge AI capabilities address these concerns.
Reduced reliance on cloud infrastructure: Edge AI can operate in remote or offline environments where internet connectivity is unreliable or unavailable. Some generative AI applications may eventually be able to run entirely locally on a device. Advanced language translation is a good example.
Apple Edge AI Advantages
Apple's advantages with integrated generative AI also carry over to Edge AI. Its vast universe of over two billion active device users provides the largest platform for implementing edge AI. Its tight software integration with new IOS releases will make it easier to release these capabilities. Finally, its expertise in advanced microprocessor development will enable it to create specific edge AI processing capabilities.
Interactive Voice Communications
Interactive voice communications is the third advanced capability of generative AI. Today, one-way voice communication is commonplace with Siri, Alexa, and others. Generative AI offers the opportunity to create more interactive communications of the type illustrated in the Knowledge Navigator video.
Let's look at an example of interactive communications using the first example of integrated generative AI.
User: Hey Siri, find the research paper I wrote 18 months ago on the potential impact of the new weight loss drugs. Then, update it with the latest research and drug trial results, add quotes from food industry executives, and include any of the latest projections.
Siri: The one on July 14, 2021, titled The Impact of New Weight Loss Drugs?
User: Was there a later version?
Siri: Yes, but under a different title.....
User: OK, use that one.
Sire: Do you want me to update the research and trial results using the previous table format or a new one?
User: A new one, please.
Siri: How many food industry quotes do you want me to include?
User: Actually, I should review them and choose, so please list them all in order of importance, and I'll delete the ones I don't want to use.
Siri: Here it is.
As you can see, the interactive dialog differs significantly from the one-way inquiries today. It is very similar to the Knowledge Navigator video.
Apple has few significant competitive advantages in interactive voice communications other than it could process some of this on the device. Microsoft and Google could and likely will provide similar capabilities.
How Could Apple (Or Others) Release These Advanced Generative AI?
So, how will Apple and other companies release these advanced generative AI capabilities: integrated generative AI, Edge AI, and interactive voice communications? Frankly, I don't know. It has several strategic options:
They could incorporate these into their software operating systems. Similar to how it includes Siri, it could add these advanced generative AI features. It would significantly raise the capabilities across all of its products. This would enable Apple to charge a price premium for the increased hardware requirements, potentially with specialized versions of its iPhone and iPad.
Apple could introduce these advanced capabilities first into the Apple Vision Pro. The thousands of readers of my article Apple Vision Pro: Everyone Is Missing The Big Picture know that I am excited about the potential of the Vision Pro. These advanced generative AI capabilities are a natural fit for the Vision Pro. It also enables Apple to justify a premium price for the Vision Pro and make it even more distinctive.
It could be introduced on an entirely new device similar to an updated version of the Knowledge Navigator. It could be a much larger desktop device like the Knowledge Navigator, replacing a portion of the user's desk. I've always been interested in such a product. According to the United States Patent and Trademark Office (USPTO), the trademark on the term "Knowledge Navigator" is owned by Apple Inc. The trademark was registered on August 25, 1987.
It could be a new specialized application. This could make sense since it may only be attractive to more advanced users initially. It would likely be a subscription-based app that could generate a significant new revenue stream.