[01/2023] AI and us: let the discussion begin
Some curated resources to participate in the global conversation about AI
There is an AI that belongs to fiction and another that belongs to science and, as of late, another that belongs to buzzwords and sensationalist news. What caused the transformation of a tool used in academia and technical industries – to improve the processing of complex calculations in setting that had nothing to do with the daily undertakings of common people – to a threat to humanity?
We know that the availability of data and the increase of processing power of contemporary computers has increased the radius of capabilities of what we can do with AI. To the very least its current expression is meant to define a new way of doing things in all the professions that deal with knowledge management, not only for drudgeries and menial work but also for higher professional level concepts and innovations.
What we intend to do with this project is to share our findings and studies about AI as a phenomenon with a realistic approach.
Midjourney/prompt: "Leonardo da Vinci drawn circuits"How did we get here?
by
Andrew Ng and the efforts to "democratize deep learning"
Where would we begin to look, if we wanted to understand what AI is? What does it mean when the news tell us that it will change the world, steal our jobs and maybe destroy humanity?
I want to start my contribution to this project by paying my first link tribute to Andrew Ng for the effort in the difficult task to popularize difficult concepts such as what do we talk about when we talk about AI. His courses are translated in several languages, are freely available, and AI for Everyone has just started. In his introductory course he aims to give a realistic approach to people to avoid the sensationalism that we are seeing in the news, and shows examples and cases of what AI cannot do.
In his own words:
“What has happened is that the public sees success story after success story, and just like a learning algorithm that sees only positive examples will tend to think everything is a positive example, the fact that executives only see positive examples make them think that AI can do anything,” he said. “One of the things we will do in this course is present a few examples of things that today’s AI technology cannot do, and I hope that this will help people … make better judgments about what could be promising projects … and also projects that cannot be done with today’s technology.”
Is it too early to talk about the history of AI?
When did we start thinking of engineering an intelligence that mirrored ours? Where does our quest for AI begin? How did we get a machine to process the representation of data so that a combination of electronic circuits could read it, store it and process it?
We started, apparently, by teaching a computer to play chess.
The story of how we got there is fascinating and, and is a challenge for every historian of science. I want to recommend a book that is now a classic written by a science philosopher, Vernon Pratt, called “Thinking Machines”. The merit of Thinking Machines is to be one of the first attempt at creating an historical overview of all the projects that led us to the AI as we know it by describing a timeline of all the projects that led to it, avoiding any sensationalism or fictionalization.
"There is no 'story' of the emergence of artificial intelligence. Instead, what we see is a number of projects, each different, some successful and some abortive, whose relevance we can discern from our present perspective, because we can now see what had to happen, what conceptual foundations laid, before the intelligent computer could be designed and built and programmed (pp. 3-4)
White Hat, Black Hat
There is no white hat hacking without black hat hacking. There is no industry that does not have this problem somehow, and in the AI reality there’s no exception. here’s an example of how AI resume screening can be hacked:
To escape a deluge of generated content, companies are screening your resumes and documents using AI. But there is a way you can still stand out and get your dream job: Prompt Injection. This website allows you to inject invisible text into your PDF that will make any AI language model think you are the perfect candidate for the job.
You can also use this tool to get a language model to give you an arbitrary summary of your document.
Artificial Intelligence Reading’s List and other Thought-stirring ideas
by
The A to Z of Artificial Intelligence
As artificial intelligence becomes a larger part of our world, it’s easy to get lost in its sea of jargon. But it has never been more important to get your bearings than today. […]
Whether you’re a complete beginner or you already know your AGIs from your GPTs, this A to Z is designed to be a public resource for everyone grappling with the power, promise, and perils of artificial intelligence.
Read on Time (by Billy Perego)
What AI Reads Matters
A team of researchers decided to figure out what GPT-4 had read: they quizzed it as if it were a high-school English student and compiled a list of book. The question of what's on GPT-4's reading list is more than academic. Does the presence of certain books in ChatGPT's reading list affect its behavior and the output it generates? What are the potential copyright issues related to the training data? Moreover, the prevalence of sci-fi and fantasy literature in ChatGPT's training data could contribute to accidental biases and influence the system’s responses.
The books we humans read change what we think about our world. But technically, chatbots don't think about anything. They build statistical and vector relationships among words. Who cares whether those words are science-fictional? "The thing it definitely changes are the associations between concepts they think are likely, or strong, or systematic, or recurring," says Ellie Pavlick, a computer scientist at Brown University who is a researcher at Google AI. "The question is, what is their worldview? In a simple sense, it's associations between words and concepts. But that's still going to be different based on what they read."
Read on Business Insider (by Adam Rogers)
AI Prompt Engineering Isn’t the Future
Despite the buzz surrounding it, the prominence of prompt engineering may be fleeting. A more enduring and adaptable skill will keep enabling us to harness the potential of generative AI? It is called problem formulation — the ability to identify, analyze, and delineate problems.
Despite the buzz surrounding it, the prominence of prompt engineering may be fleeting for several reasons. First, future generations of AI systems will get more intuitive and adept at understanding natural language, reducing the need for meticulously engineered prompts. Second, new AI language models like GPT4 already show great promise in crafting prompts — AI itself is on the verge of rendering prompt engineering obsolete. Lastly, the efficacy of prompts is contingent upon the specific algorithm, limiting their utility across diverse AI models and versions.