AI How It Works

Why Settle For a Gatekeeper to Search, When AI Could Be Something Exceptional?

OmniIndex’s Matthew Bain asks “Do we really want this amazingly powerful AI tech to be focussed on advertising revenue, or do we want more?”

Intro

It is impossible to escape AI at the moment. And we all know why. ChatGPT has caused massive disruption not just in the tech world, but across all industries with many stating that “our relationship with computers will never look the same” again. (For example Freund in Forbes)

However, beneath the hype of quick answers and impressively self-generated content fuelling this disruptive momentum, there’s some truly incredible tech. Tech that has the genuine power to “perform exceptionally” at specific tasks, as opposed to this Generative AI which creates content “often less consistent compared to that produced by human experts.” (OpenAI, 2023-02-10, ‘Which is better? Generative AI, or Narrow AI’)

This article looks at the ‘exceptional’ power of Narrow AI, and why this needs to be the future as opposed to settling for AI’s current use: a shortcut to ‘less consistent’ content. 

Quick Definitions

AI: Artificial Intelligence (AI) refers to the development of computer systems that can perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, and language translation. (OpenAI, 2023-02-10, ‘AI in a sentence?’)

Generative AI: Generative AI is a subfield of Artificial Intelligence that focuses on the creation of new, unique outputs, such as images, music, or text, by training AI models on large datasets. (OpenAI, 2023-02-10, ‘Generative AI in a sentence?’)

Narrow AI: Narrow AI refers to artificial intelligence systems that are specifically designed to perform one particular task with a high level of accuracy, such as image recognition or speech recognition. (OpenAI, 2023-02-10, ‘Narrow AI in a sentence?’)

The Problem

Generative AI is undeniably powerful. However, it is deeply flawed as it essentially just regurgitates what it has digested. As a result, it “can be corrupted by prejudice and bigotry if it is trained on biased data, as it may then generate outputs that reflect and reinforce those biases.” Indeed, “it is possible for a malicious third-party to try to manipulate the inputs to the model in an attempt to elicit biased or harmful responses.” (OpenAI, 2023-02-10, ‘Could generative AI be corrupted by prejudice and bigotry?’)

The accuracy and authenticity of its results can also be easily corrupted by money. This is because the measure of ‘success’ heavily influences the output of the engine. For example, if the measure of success is based on the value of the revenue associated with the results, and not their accuracy and relevance, then instead of results being the most accurate and best possible, they will simply be the ones with the best bid behind them. 

In the words of ChatGPT: 

“It is possible for a generative AI tool like myself to be used to provide answers that are influenced by advertising or marketing interests. For example, if the training data for the AI model includes information from sources that are biased towards certain products or services, the AI’s responses may reflect that bias.” (OpenAI, 2023-02-10)

My concern is that if we continue using AI in the same way as the disruptive hype expects and demands, then it will produce content ‘less consistent compared to that produced by human experts’ and will become skewed towards bought answers that generate revenue instead of providing the most useful and accurate responses. 

So my question is this: 

“Do we really want this amazingly powerful AI tech to be focussed on advertising revenue, or do we want more?”

We Want More!

If you said you want more, and I really hope that you did, then my solution is probably already obvious from the bias built into this article from the start: Narrow AI. This is because it is the type of AI that can actually outperform humans and “perform exceptionally” at its specified and dedicated task. For example, while Generative AI like ChatGPT can tell me what tasks Narrow AI can do better, Narrow AI can actually do them. Tasks such as:

  1. Perform highly specialized tasks: Narrow AI systems are designed to excel at specific tasks within a narrow domain, such as playing a game or identifying objects in images. They can perform these tasks with greater accuracy and speed than Generative AI systems.
  2. Scale to large amounts of data: Narrow AI systems can handle large amounts of data within their narrow domain of expertise. In contrast, Generative AI systems may struggle to process large amounts of data due to the complexity of their models.
  3. Provide more interpretable results: Narrow AI systems often generate results that are easier to interpret and understand than those produced by Generative AI systems. This is because Narrow AI systems are designed to focus on a specific task, while Generative AI systems may generate more complex and abstract outputs. (OpenAI, 2023-02-14, ‘3 examples of tasks Narrow AI could do better than Generative AI.’)

One specific example of Narrow AI performing a task which neither a human nor Generative AI could do, is generating insights with near instant speed from fully encrypted data having been only ‘trained’ by uploading the corresponding dictionaries and resources to that specific industry moments before…

Use Case: The Narrow AI powering The OmniIndex Data Platform

The OmniIndex Platform enables the same analytics available on plaintext data to be performed on encrypted data. Data is stored in the user’s own blockchain and is protected from unwanted third-party eyes and Ransomware attacks. This means customers can add their most sensitive and confidential data to collaboration, productivity, and analytics tools with no risk of exposure.

Hidden within the depths of this data platform, is the OmniIndex Narrow AI engine.

The AI’s job is to enable authenticated and authorized users to access and use the Fully Homomorphically Encrypted data stored in the user’s blockchain. Because the processing power and libraries inputted into the AI are dedicated to this one task, it performs it exceptionally. And what’s more, it does a job impossible for a human to do. 

Put simply: OmniIndex’s Narrow AI performs a specific task better than any person possibly could, as opposed to being a gatekeeper to search results or a shortcut to inferior content production. 

To learn more about how the OmniIndex narrow AI works and how AI is powering OmniIndex’s FHE Blockchain, read our Q&A with OmniIndex CEO Simon Bain.