#### Greg Bernhardt

We asked our advisors “What technology or scientific discovery will be the most revolutionary within the next 20 years?” Here are their responses…
Demystifier
“Machine learning will be ubiquitous, just as applications of traditional computer algorithms are ubiquitous now.”
Ygggdrasil
The ability for scientists to create “artificial gametes”—functional human sperm and eggs cells produced in the lab—could fundamentally change society. The idea would be that a doctor could take a skin cell from a patient, convert those skin cells into stem cells in the lab via cellular reprogramming, and culture the stem cells in a dish to coax them to differentiate into germ cells capable of generating sperm or eggs. Scientists...

Last edited:
Related General Discussion News on Phys.org

#### Drakkith

Staff Emeritus
2018 Award
The ability for scientists to create “artificial gametes”—functional human sperm and eggs cells produced in the lab—could fundamentally change society. The idea would be that a doctor could take a skin cell from a patient, convert those skin cells into stem cells in the lab via cellular reprogramming, and culture the stem cells in a dish to coax them to differentiate into germ cells capable of generating sperm or eggs.
Interesting. This is the first I've ever heard of this idea.
Thanks, @Ygggdrasil.

#### Greg Bernhardt

I'll throw in for advanced robotics in manufacturing.

#### gleem

I believe that AI will be the most revolutionary in the next 20 years. It will greatly impact society and even science and will be disruptive causing us to rethink many traditional ideas and values.

Many experts only see specific task AI which has made rather significant advances in recent years but expect slow progress in the next ten to twenty years. They point to the fact that what we are seeing today is based on decade old approaches using costly computer/cloud resources for the noted progress as well as using systems based on technology designed for other application like graphic processors. These systems are physically elaborate and costly both in money and power requirements. We are seeing new algorithms and approaches to AI.

But I believe that the technology will accelerate in the coming years with the use of more flexible computing hardware as field programmable gated arrays in the near term and the development of applications specific instruction set processors. While AI today is task specific i.e, doing one thing well, future AI will integrate many application specific processors to accomplish complex tasks requiring the coordination of various forms of information acquisition and processing just like our own brain in which various areas have specific functions and coordinate with other areas to work on varied tasks.

Last edited by a moderator:

#### MathematicalPhysicist

Gold Member
a personal Quantum computer, and quantum mac.

PQC.
Will such a computer be able to prove the claymath conjectures with suitable programme?

#### Redbelly98

Staff Emeritus
Homework Helper
Can't say that this will happen in the next 20 years, but some day: Discovering the particle or other explanation responsible for dark matter.

Gold Member
2018 Award

#### hilbert2

Gold Member
Electrical brain stimulation can improve memory and learning performance:

#### Rive

Well, I think real revolution will not come from a big discovery or new technology, but from change of attitude regarding human 'enhancement' (bio-) technologies. I expect ~ a decade long resistance due some problems around ethic and/or religion, followed by a sudden change like a dam bursting. Slowed ageing, longer active lifetime, disappearance of some diseases, maybe some regenerative technologies, enhanced performance in some areas...

#### DoctorSatori

A relational philosophy of science that gives us a better idea about what scientific problems are worth solving.

#### BWV

physics had its heyday last century with meager prospects of great discoveries over a 20 year time horizon (what great revolutionary discoveries has physics made since 1999?) Maybe there will be some great new materials science / condensed matter stuff? but whatever physics does will be dwarfed by biology over the next few decades

Last edited:

#### PAllen

Definitely not controlled fusion, because that is always 50 years in the future.

BWV

#### haael

A dream control device.

A thing that allows you to induce dreams in humans.

This would obsolete cinema, games and maybe even dating.

We will be transformed into a society of dreamers Matrix-style, just as today we are addicted to mobile phones. Sure that will be a social change!

#### hutchphd

Definitely not controlled fusion, because that is always 50 years in the future.
No I'm sorry but it is always 20 years in the future....

#### Ryan_m_b

Staff Emeritus
a personal Quantum computer, and quantum mac.

PQC.
Will such a computer be able to prove the claymath conjectures with suitable programme?
Affordable and reliable quantum computers would be revolutionary in chemistry, materials, biotech and medicine (at least). Simulating molecules and their interactions is very hard and relies on a lot of approximations, consequently even the best computational chemistry work is paired with rounds of practical experiments. A quantum computer would be able to simulate chemistry without these approximations, the potential for that is vast. Anything from searching for more efficient photosynthetic molecules (and efficient pathways to synthesise them) to hunting for new medical candidates by simulating potential targets and how to inhibit them.

#### MathematicalPhysicist

Gold Member
Affordable and reliable quantum computers would be revolutionary in chemistry, materials, biotech and medicine (at least). Simulating molecules and their interactions is very hard and relies on a lot of approximations, consequently even the best computational chemistry work is paired with rounds of practical experiments. A quantum computer would be able to simulate chemistry without these approximations, the potential for that is vast. Anything from searching for more efficient photosynthetic molecules (and efficient pathways to synthesise them) to hunting for new medical candidates by simulating potential targets and how to inhibit them.

#### Ryan_m_b

Staff Emeritus
Sorry I don’t know the answer so wasn’t attempting to answer it, instead offering why I think QCs could be an incredibly revolutionary technology in the coming decades.

#### fresh_42

Mentor
2018 Award
Will such a computer be able to prove the claymath conjectures with suitable programme?
No.

Just a few examples:
• $P\neq NP\, : \,$ We need a proof for something cannot exist. Computer capacities won't help.
• $ERH\, , \,\text{Goldbach}\, : \,$ We already checked $10^{10}$ zeroes. Some more won't help.
• $\text{Birch-Swinnerton-Dyer}\, : \,$ same as ERH
• $\text{Navier-Stokes}\, : \,$ A computer can give a numerical approximation and so does a pipe, but this doesn't solve the problem.
• $\text{Yang-Mills}\, : \,$ A quantum computer might be helpful to deal with some exceptional groups, after a general solution has been found for the regular ones.

#### BWV

No.

Just a few examples:
• $P\neq NP\, : \,$ We need a proof for something cannot exist. Computer capacities won't help.
• $ERH\, , \,\text{Goldbach}\, : \,$ We already checked $10^{10}$ zeroes. Some more won't help.
• $\text{Birch-Swinnerton-Dyer}\, : \,$ same as ERH
• $\text{Navier-Stokes}\, : \,$ A computer can give a numerical approximation and so does a pipe, but this doesn't solve the problem.
• $\text{Yang-Mills}\, : \,$ A quantum computer might be helpful to deal with some exceptional groups, after a general solution has been found for the regular ones.
But that is brute force calculation, what about AI actually doing the proofs?

#### MathematicalPhysicist

Gold Member
But that is brute force calculation, what about AI actually doing the proofs?
Yes that was what I referred to when I said a QC with a suitable programme like the proof-assistants software we have nowadays, like Coq.

#### phinds

Gold Member
• medicines that are more specifically targeted to specific diseases and specific individuals.

#### fresh_42

Mentor
2018 Award
But that is brute force calculation, what about AI actually doing the proofs?
This cannot be answered as AI as well as proofs are both hypothetical. I find it difficult to encode the concept of for all in an AI, i.e. I cannot imagine that this can be done. Furthermore the number of potential proofs in the sense of a finite sequence of conclusions is so incredibly huge, that I think it needs phantasy and creativity, which again I cannot imagine could be added to an AI. But the discussion is meaningless so far. The best AI of today can just pretend to understand, so the discussion will have to be postponed a few decades.

#### atyy

This cannot be answered as AI as well as proofs are both hypothetical. I find it difficult to encode the concept of for all in an AI, i.e. I cannot imagine that this can be done. Furthermore the number of potential proofs in the sense of a finite sequence of conclusions is so incredibly huge, that I think it needs phantasy and creativity, which again I cannot imagine could be added to an AI. But the discussion is meaningless so far. The best AI of today can just pretend to understand, so the discussion will have to be postponed a few decades.
I think there is a good argument that AIs can encode the concept of "for all", since human beings have that ability and they encode the concept using neural networks (and there is arguably not much difference between biological and artificial neural networks in terms of encoding capability).

I have seen it stated that artificial neural networks are equivalent in some sense to Turing machines. https://arxiv.org/abs/1410.5401: "Moreover, it is known that RNNs are Turing-Complete (Siegelmann and Sontag, 1995), and therefore have the capacity to simulate arbitrary procedures, if properly wired."

I agree that doing creative mathematics will be hard for AIs, but that is also hard for humans. In terms of music, AIs before the current renaissance could already write music that sounded quite good for short segments, although large scale structure was probably problematic (as it generally is also for humans):
https://www.computerhistory.org/atchm/algorithmic-music-david-cope-and-emi/

Last edited:

"What Technology or Discovery Will Be the Most Revolutionary Within 20 Years? - Comments"

### Physics Forums Values

We Value Quality
• Topics based on mainstream science
• Proper English grammar and spelling
We Value Civility
• Positive and compassionate attitudes
• Patience while debating
We Value Productivity
• Disciplined to remain on-topic
• Recognition of own weaknesses
• Solo and co-op problem solving