5
$\begingroup$

Right now, Google et al. will find anything anyone has written, without regard to the accuracy or quality of the source. If anything, Google is creating intellectual bubbles for people by learning which links they are likely to click on, reinforcing their existing biases and misinformation.

Look at the success of Watson at solving tricky natural-language wordplay. Such systems will be used for medical and legal knowledge bases, and I've already started seeing advertising for that.

So imagine an advanced computing resource that sucks in all the information known to humanity (like Google) but then digests it to apply scientific reasoning to statements of fact. Suppose that curators work full time to initially set up evaluations of which sources are high or low quality in general and then in specific fields, and answer questions Multivac (named after Asimov’s fictional super-computer) comes up with as it digests things like what Cyc would do.

Eventually Multivac would start pronouncing which journals were highest quality and which rags were worst. The curators would be employed in ever higher level discussions trying to make sense of humanity.


Multivac is put on the Internet for all. It is funded by industrial use, for projects involving product design and scientific research and commercial analysis of legal and medical reports. But for personal use anyone can ask questions for free and get simple factual assessments.

If you ask something like “what is the best alloy to use in this application” you will be forwarded to the commercial division where experts will set up the precise parameters and you can purchase a detailed report.

If you ask “what medicine should I buy” it will say sorry need specific question. Ask “will medicine X help my cold” it will calmly say “No, it has no effect other than placebo and you just ruined that by finding out.” If you ask which medicine is best it will refuse to answer; evaluations are commercial and need experts to set up and produce reports not simple answers. If you ask it which medicine is strongest it will identify those that have the highest dosage of the active ingredient.

I also imagine it being able to annotate a web page, perhaps color coding statements as to accuracy with links to more details and corrected information.

My question is: how will people react to this? There are different societies in the world and different cultures and segments within that. I think about how people already know that certain expensive homeopathic products are just plain water, but rationalize them in crazy and innane ways. So just knowing is often not an issue.

But consider the different levels of what can be asked: what’s in it, what mechanism does the seller claim, or just plain does it work? As people learn to trust the results, will they go for the high-level answer and skip over the intermediate layers where the questioner’s bias would conflict?

Just like people stop being able to do arithmetic once calculators appeared, will a device people can employ to do critical thinking for them result in people less capable?

How will this affect various societies and different groups of people? What will happen to those who “resist”?

$\endgroup$
4
$\begingroup$

There's really very little that we can say is absolutely true. The issue you described with medicine, well, that's going to happen for just about EVERY SUBJECT.

Effects as I see them:

People would learn how to ask questions So you illustrated why a particular question wasn't valid, although it's a common one. Should Multivac be wildly popular, I think that, at first, it would make people into more critical thinkers because they would have to narrow things so much.

It would not be popular unless there were nothing else People use google precisely because it is easy and it does answer common questions. I can see it being used as a tool for a sub-set of professionals, scientists, or hipper-than-thou intellectuals, but the general public simply doesn't want to think about such things when asking a question.

Media/News Use It also might become a tool for the media. Just like google, they won't double check it, so if anything happens to be wrong, it will be reported that way.

Garbage in Garbage Out Although you have checks and balances inherent in this system, the info comes from us falliable humans, who do tend to taint data in new and creative ways. That is another reason why the useful data will be limited.

Confirmation Bias will Absolutely Survive this Don't worry about all those folks who believe that water that someone dipped silver in will cure them of everything. They will still continue to believe in whatever they believe in. But it won't just be them. Every human on the planet tends to do this. And they will continue to do so, long after your machine is proven correct, over and over again.

People in Power Would Manipulate the Data Recall what I said about garbage in garbage out? Well, if this does remain pristine in the first few years as far as data entered, and makes headlines as a source of unbiased fact, well...if I wanted the country or the world to believe something and I were in power or in business, if I were unscrupulous, I might feed the AI bad data to get what I wanted.

Very little would change. Because people are terrible.

$\endgroup$
3
$\begingroup$

There are too many questions are simply not susceptible to scientific reasoning and inquiry. particularly there are too many beliefs people will cling to irrespective of what Multivac says is or isn't true. Indeed many people will harden their beliefs in the face of contradiction. This is already a well known human failing.

Knowledge is never enough. Critical thinking, even if carried out by supposedly infallible source will be subject to doubt and disbelief. If people burned books, they will soon want to mangle Multivac.

Will a device capable of doing their critical thinking for them result in less capable people? No. People are already insufficiently capable of doing critical thinking for themselves; this won't make them their less capable. Multivac will only be source of much disagreement and dissent.

The other great limiting factor is simply we do not have the answers to many questions. These questions about the unknown. The things science doesn't know the answer about. Multivac would be a curse for scientific research. It would reinforce what we don't know and the prejudices of those who think they do.

Remember Max Planck said "science advances one funeral at a time." Wrong and erroneous ideas are shed piecemeal as those who believed in them shuffle off this mortal coil. Multivac will inevitably contain much knowledge that is plain wrong. There will be many areas where what we think we know is shot through with contradictions.

Humanity will be as confused and uncertain as it was before. Multivac will be too.

$\endgroup$
2
$\begingroup$

If only simple questions are answered, the situation is not significantly different from today, where information of quality exist and is accessible. Only the cost (in time) and the scale would change.

Not more than today will people stop to believe whatever they wish be the truth (even when proven false) : homeopathy, god, little green men in the White Office, etc.

People may rely on this more and more over time and, like mentioned, get rid of doing the basic reasoning by themselves. In the past, tools replaced muscular force, machines replaced repetitive labor, computers started doing arithmetic. There is no need such creation will not replace basic reasoning.

What is interesting is that even the question states that this will make us less capable, which sounds biased. At different point in time, the same was said for the fire that make us weak and less capable of resisting cold, for the tools that reduced our muscular mass, and of course for machines and computers.

Is that bad ? Or is that only our fear speaking ? Maybe such change will open a new era where mankind will free itself from basic activities and reach another step in its evolution, like for example a global consciousness, or full understanding of reality. No opinion here, but the simple remark that we see this through the lens of the present.

$\endgroup$
  • $\begingroup$ Could you include a link for the white office? I only find literal results on Google, even if I add french as a keyword. $\endgroup$ – JDługosz Aug 29 '16 at 6:32
1
$\begingroup$

Multivac will have a huge problem: much of the knowledge that it is using to make pronouncements are just wrong.

This article touches on the issue of papers being published which tout certain results from experiments, which turn out to be either wrong or not reproducible. This has been reported in numerous journals, but the current "publish or perish" environment results in a tidal wave of papers listing experiments with dubious or statistically insignificant results, but far too many to have other labs redo to check for accuracy.

And then there is scientific fraud. A paper was published and widely cited suggesting that people could change their attitudes about homosexual marriage if people canvassed them door to door and engaged then on their beliefs. Fortunately for science, someone noticed statistical anomalies in the research and it eventually came out that the entire "experiment" was simply made up. While this is good in the sense that it shows how science self corrects, the paper was widely circulated and cited, including in media and social media, planting a very wrong idea in people's heads. Policy makers may be influenced by stuff like that, and it is impossible to know how much is incorrect or even fraudulent in the "scientific" backing for many programs, laws or regulations. Since the background research is corrupt, Multivac won't be able to make accurate pronouncements. (Indeed, Multivac would be better served by using its power to identify and isolate questionable research and get people to redo the experiments).

Then of course there are things which cannot be scientifically proven. Economics is a descriptive science, since it cannot run experiments on nations and the global economy, but only draw inferences from small scale experiments and large scale observations. I can confidently predict the economy will grow if taxes are cut, but there is no possible way to say by how much or in which sectors.

Going beyond that, there are questions which have no clear answers because of chaos theory; the question is about something embedded in an adaptive, non linear environment where inputs and outputs are not spatially, linearly or temporally connected (input A does not always result in output B). Climate, ecosystems, military strategy and the stockmarket are examples of this sort of system.

And finally, as many people noted, there are questions which are essentially opinion based, or which cannot be framed accurately enough to narrow down one particular answer (much like Worldbuilding Stack Exchange!)

Given the rather severe limitations, people might decide that Multivac is not an accurate tool or one worth consulting, except for very narrowly bounded subjects. For the mass of people, it will be essentially useless, and become the realm of researchers who need very specific and narrowly bounded questions answered.

$\endgroup$
0
$\begingroup$

In the presence of experts people get lazy. Let them handle it, they know how to do it/they know better.

Expect people to get lazy, to not bother anymore, become hedonic. Even if many people initially resist, that will erode to a small group of 'fanatics' that will keep resisting based on faith.

There are many science fictions books that follow this pattern: the people's needs are taken care of by technological means, and they fall into apathy. Of course, those books have resistance fighters who eventually overcome all this, otherwise the books would be uninteresting. But I doubt that is what really would happen. We would be far too happy that many of those 'difficult issues' can now be taken care of by Multivac.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.