Key Messages
Exploring How We Got To The Race For True Ai, And What Might Happen Next
Genius Makers takes a look at the exciting field of artificial intelligence, starting in 1968 with the famous movie, 2001: A Space Odyssey.
In the film, HAL is a supercomputer that has its own mind and agenda, yet for many decades, such advanced technology was still largely science fiction.
Today however, we are seeing significant advancements towards AI in many different sectors thanks to government research initiatives as well as many entrepreneurs and professor from around the world.
This book thoroughly explores how this all came to be over the past few decades and where current trends could lead us in the future.
Based on interview with specialist at major companies like Google, Microsoft and OpenAI and more detailed research, it shows just how much moving forward we’ve already made with AI related technologies already integrated into our lives.
Besides learning why you must trust what you see online, readers will also be able to learn what makes their brain superior to a computer and how one book ended up causing an “AI winter.”
The Rise Of Artificial Intelligence: From Skepticism To Research And Development
When artificial intelligence was first introduced, it was met with a great deal of skepticism.
One particular example of early research into AI was Frank Rosenblatt’s Perceptron, which he presented in 1958 at the United States Weather Bureau in Washington D.C.
This machine could identify black squares on cards, showing a basic level of understanding which was seen as an early precursor to AI.
Despite this, at the time it still faced criticism for being viewed as merely a novelty.
Further research into AI and neural networks occurred in 1960 with the Mark I computer, which analyzed paper cards to distinguish between letters such as A B or C.
This process involved using a series of calculations to guess which letter it had been given and humans then marking the results as correct or incorrect so that the machine could learn more accurately on subsequent attempts.
Although some scientists compared this process to those of the human brain (called connectionism), others like MIT computer scientist Marvin Minsky were less supportive and criticized its limitations on complex problems.
This resulted in few institutions funding neural network research over the next two decades and essentially created what is now called an “AI winter”.
Ultimately, this significant period of skepticism towards AI has led where we are today by helping us understand its potential and its limits alike.
Geoff Hinton’S Outsider Status Leads To A Revolutionary Deep Learning Discovery
Deep learning made neural networks technology‘s newest tool in recent years.
This advancement started when Geoff Hinton and Li Deng met at NIPS, an AI conference in British Columbia, Canada.
Hinton suggested that deep learning could potentially outperform a conventional approach to computer science, which piqued Deng’s interest enough for them to start a project together.
Over 2009, they worked on creating a program that used machine learning models to analyze hundreds of hours of sound recordings.
To aid their research, they ran the program on special GPU processing chips usually reserved for gaming computers.
After processing the data, it was clear that this method had great accuracy–in some cases correctly picking out words with error rates as low as 18%.
This was all the proof needed for neural networks to become tech’s new favorite toy.
Companies like Google jumped on board as tech giants vying to purchase AI startups and further invest into this revolutionary technology, making sure it only got better from there–and better it did get!
With self-driving cars relying on navigation systems fed by Deep Learning data analysis and image search being more accurate than ever before taking advantage of the power of Deep Learning herself we’re seeing just how far Neural Networks have come since Hinton first decided to use them in his research efforts so many years ago.
Silicon Valley Companies Invest Big In Ai Research To Gain A Competitive Edge
The tech giants of Silicon Valley were all recognizing the potential of artificial intelligence (AI).
They saw AI as a path to enormous profit, and the ones with the most money—like Facebook, Apple, and Google—wanted to get ahead of the pack.
So they did what they do best: throw money at it.
They began investing heavily in AI research, spending millions on recruiting AI specialists from universities.
Places like NYU’s deep learning lab were particularly popular targets for these job offers—offers that were typically capped off by a personal appeal from the CEO himself!
Deep learning and neural networks weren’t quite mainstream yet, but the ambitious tech entrepreneurs understood their potential and long-term value.
Facebook could use neuron nets to parse massive server data to display targeted ads and power chat bots.
For example, Google was working on self-driving cars that used Street View data as training material.
Microsoft was attempting even further reaches with its AI projects, such as making servers more energy efficient.
Ultimately, Silicon Valley invested in AI out of simple ambition: no one wanted to come in second place when it came to developing innovative–and profitable–AI technology directions.
Despite some warnings about unpredictability and risk posed by superintelligent machines, no one slowed down their roll towards innovation.
The Power Of Neural Networks: How Machines Continue To Surpass Human Intelligence
For years, scientists have been exploring the power of neural networks to unlock the potential of computing.
Their research bore fruit in 2015 when Google’s AlphaGo AI took on the reigning Go champion and won five games in a row.
This was the first time a computer ever beat a human player at this game, and it marked a major turning point in machine learning and artificial intelligence.
The achievements of AlphaGo indicated that neural networks could outdo humans in many fields.
With access to faster computers and bigger data sets, these networks can be trained to analyze a vast array of information, from millions of Go games to X-rays, CAT scans, MRIs and more.
As demonstrated by Varun Gulshan and Lily Pengs work with diabetic retinopathy diagnostics, neural nets could soon become essential tools for accurate medical diagnoses across countries like India where there may not be enough doctors to evaluate everyone.
In short, neural networks are truly revolutionary technology that have wide-ranging implications for our future — they have the potential to outdo humans in many fields.
Ai Has The Potential To Distort Reality, Raise Ethical Concerns, And Propagate Inaccuracies
As AI technology progresses and becomes more advanced, there is a growing potential for it to distort our view of reality.
This problem is becoming increasingly concerning as many AI applications are based on data that has already been skewed and distorted in favor of certain demographics – namely white males.
The emergence of generative adversarial networks (GANs) further compounds this issue by making the generation of realistic images easier than ever before.
These deep fakes, when used unethically and maliciously, can be extremely damaging to both individuals and society by spreading misinformation.
Ultimately, sophisticated AI has the potential to open up a world of possibilities – however it is hugely important that these technologies are regulated responsibly, fairly and ethically.
Otherwise they could create an even greater divide between those who have access to these resources and those who don’t.
Ai Can Be Abused For Political Purposes, But The Fight To Stop It Is Ongoing
AI is increasingly being used by the government and private organizations to carry out tasks that are not only beneficial but also have potential for dangerous misuse.
In 2017, The US Department of Defense partnered with Clarifai’s AI research & start-up to build a neural network capable of identifying people, vehicles and buildings–especially for desert locations.
While this instance didn’t come to fruition, similar moves have been made in countries like China who’ve invested heavy funds in developing AI systems for military purposes.
Additionally, Facebook was criticized in 2016 for having their user data harvested by Cambridge Analytica which led to the creation of many misleading campaign ads.
Beyond this, Facebook has since recognized the need to use AI to moderate content on their platform but even advanced neural networks can struggle with complex nuances of politics.
Overall, it cannot be denied that AI has been and is being misused as a political tool.
This can range from helping governments create weapons technologies to creating false information and campaigns designed to warp public opinion — all of which shows just how easy it is for people (or entities) with malicious intent to exploit advancements in technology whenever they please.
Understanding Human Language Through Context-Specific Neural Networks
At Google’s I/O conference in Mountain View, California the crowd was wowed by the demonstration of its newest AI innovation, the Google Assistant.
It may have been able to make phone calls and reservations that were nearly indistinguishable from a human voice, but New York University psychologist Gary Marcus tips his hat off to this accomplishment as he rolls his eyes.
Marcus is dubious about the capabilities of machine learning – particularly when it comes to understanding complex language tasks and discourse.
He believes there is an inherent difference between human intelligence and neural networks based on nativism theory – that humans have an innate ability that deep learning cannot match.
To demonstrate their point, a nativist would argue that a child can be taught to identify a single animal after just seeing one or two examples, while neural networks require millions of images before they can obtain similar results.
It’s why neural net AI hasn’t shown as much progress when it comes to hard situations involving language.
When compared to humans, AI can still push through conversation but they lack the sophistication and comprehension needed for more technically demanding tasks like understanding jokes or metaphors.
Nevertheless, teams at Google and OpenAI are seeking ways around this problem via universal language modeling which uses context-specific approach latched onto by neural nets in order to understand speech better.
This has already yielded some accomplishments but if AI will ever understand jokes remains unknown for now.
We’re On The Path To Artificial General Intelligence But We Don’t Know Where It Will Lead
Researchers continue to push the boundaries of what is possible with artificial intelligence (AI).
We have already seen incredible milestones in the advancement of AI, but some of the brightest minds in Silicon Valley believe that further progress could take us beyond our wildest imaginations.
OpenAI has made developing AGI, or artificial general intelligence, an explicit goal of the company.
Microsoft responded by investing more than one billion dollars into their research team.
Companies like Google, Intel and Nvidia are all racing to develop processing chips specifically designed for neural networks in order to achieve new breakthroughs.
On the other hand, Geoff Hinton’s research focuses on capsule networks which more closely mimic the human brain in structure and function.
No matter what new ideas come about in the future, it’s clear that researchers are determined to keep pushing beyond current limits so that AI can benefit society as a whole – from increasing automation capabilities to making predictions and decisions faster than ever before.
Wrap Up
The final summary of Genius Makers is that recent advances in artificial technology have inspired awe, concern, and debate.
A lot of what we now perceive as AI is grounded in neural network models, which execute complicated computations to recognize patterns from massive volumes of information.
Governments and corporations alike are making use of this technology for a range of objectives from optimizing picture search results to piloting automated aircraft.
Where AI research will take us in the future is uncertain but some are hopeful it will keep bringing about innovative products and solutions.
