Can machines invent things without human help? These AI examples show that the answer is ‘yes’

The question of whether artificial intelligence (AI) can be invented is almost 200 years old, going back to the early days of computers. Victorian mathematician Ada Lovelace wrote the first computer program. In doing so, she wondered about the limits of what computers could do.

1843 Lovelace wroteregarding what is supposed to be common purpose first programmable computer:

“The Analytical Engine doesn’t expect to produce anything. It can do anything we know how to command it to do. It can obey analysis; but it’s not capable of it. predict any relationship or analytical truth. Its job is to assist us in making available what we are already familiar with.”

And this assertion has haunted the field of AI ever since. As many reviewers will note, computers just do what we tell them to do.

A century after Lovelace opposed the invention of machines, Alan Turing, one of the inventors of the electronic computer, returned to the subject. In 1950 Turing wrote what is generally considered the first book Scientific articles on AI. In it, he attempted to refute Lovelace’s objections:

“Who can be sure that the ‘initial work’ he did was not simply the growth of the seed sown in him by teaching, or the effect of following general principles well known. A better objection variant says that machines can never ‘surprise us’. This statement is a more direct challenge and can be met directly. Machines surprise me with great frequency.”

This has not changed. Today, machines surprise us more and more. Just take OpenAI’s new ChatGPT chatbot as an example. Indeed, there is growing evidence that AI can help people invent—and in some cases, may even be credited as the inventor himself.

Things AI Invented

The question of whether machines can be invented has now begun to cause controversy around the world. Stephen Thaler, co-founder of Scentient.ai, submitted Patent application for two inventions where a named neural network is the sole inventor.

These applications have been rejected in almost every jurisdiction, mainly on the legal basis that the inventor must be human. But no legal case to date has tested Thaler’s claim that the computer was really the sole inventor.

In a post published today in Nature Machine Intelligence, we examine Thaler’s claim. While we discovered many technical reasons why computers weren’t the only inventors in this case, we also have a long history of AI being used to help people develop. invented—and in some cases self-invented. These are just some examples.

3D . circuit

In the 1980s, AI researcher Douglas Lenat of erisko system (Europe is Greek for “I discover”) invented some novel 3D circuits. A provisional US patent application was even filed for one of these.

weird plane

Starting in the 1990s, computer scientist John Koza applied genetic programming to invent a number of new devices, including some rather odd radio antennas that resembled paperclips. Bend. One of these antennas could be the first AI invention in space, because it flies on NASA’s ST5 spacecraft.

Toothbrush

Although it is not a better mousetrapin 1998, the Oral-B CrossAction Toothbrush was invented by the aforementioned Stephen Thaler during a brainstorming session. session with a neural network.

antibiotics

More recently, researchers at the Massachusetts Institute of Technology used a neural network arrive determine Halicin—a powerful new antibiotic compound. Halicin is named after HAL, the AI ​​computer made famous in Arthur C. Clarke’s 2001: A Space Odyssey. Many companies with billions of dollars in funding are using AI-based strategies for drug discovery and development.

Looks like the AI ​​invention is still here.

But is AI an ‘invention’ or helping humans invent?

The abstract idea behind how AI programs can invent is relatively simple. You define several spaces of concepts and programs that explore this space. Space is often very large, possibly even infinite. Therefore, considerable effort must be invested in determining whether a part of the space is worth further exploration, as well as to confirm any promise of a new concept.

For example, the space of concepts could be all possible ways to bend a straight plane. The challenge is to figure out which of the myriad ways has the best electromagnetic properties.

We asked Jurassic chatbot-1, a cousin of ChatGPT, to issue a patent like one of Thaler’s patent applications. Here’s what we have:

“PVC, latex or silicone rubber gloves, especially disposable gloves. The present invention provides gloves with a flexible clamp made up of a fractal pattern. The grip is made up of a continuous fractal pattern. . The flexible clamp is strong and stiff enough to perform its intended function.”

To see if the idea was truly original or at least unpatented, we searched the US Patent and Trademark Office’s online database and found nothing. which patent contains the words “glove” and “fracture”. Therefore, it is possible that a glove with a flexible fractal grip style could be patented.

Importantly, this idea is generated independently by the computer without human help or prompting.

So where did this leave us?

Just as AI is changing other aspects of our lives, it looks like it will soon change the way we invent. We need to think carefully about how the innovation system adapts to these changes. AI can reduce the time and cost associated with invention, while increasing the technical depth of inventions.

Do we need a new form of intellectual property to protect AI-generated inventions? Or will patent offices be flooded with new patent applications invented with the help of (or by) AI?

Put on your fractal gloves and wait for the surprise!

This post was reposted from Conversation under a Creative Commons license. Read original article.