This AI pollster has predicted the chances of a win by Biden or Trump
Category: News & Politics
Via: buzz-of-the-orient • 4 years ago • 29 commentsBy: Nicole Bogart CTVNews (Canada Television News)
BUZZ NOTE: There is a relevant news video with this seed which can be accessed by clicking the 'SEEDED CONTENT" link just below this message, which will take you to the original source article.
S E E D E D C O N T E N T
This AI pollster has predicted the chances of a win by Biden or Trump
This AI pollster has predicted the chances of a win by Biden or Trump
TORONTO -- Polly Pollster, the artificial intelligence pollster that correctly predicted the Liberal’s minority win in the 2019 Canadian election, has found that U.S. President Donald Trump has a slim chance of winning the White House this time around.
Using social media data to predict how people’s online behavior and sentiment predicts their real-world actions, the AI pollster predicts that Trump has a less than eight per cent chance at winning the Nov. 3 election.
In fact, if Polly’s predictions are correct, the president’s only hope lies in a credible treatment or vaccine for COVID-19.
“Vice President Biden is far ahead of Trump,” Erin Kelly, CEO of Advanced Symbolics Inc., told CTV News Channel Saturday.
“The one thing that Polly did see, based on the two debates, is if by chance there is a credible vaccine or treatment for COVID in the next week – and President Trump alluded that he might have something – that would really be his only hope at this point.”
Polly is not a traditional pollster. The artificial intelligence system, built by the Ottawa-based startup, scrapes public data from social media networks to predict election outcomes.
“She follows the same methodology as traditional pollsters in that she gets a representative sample of the population, but, in this case, she’s doing it on social media,” Kelly explained.
The result? A much bigger sample size, Kelly says, along with the unique ability to eliminate bias.
“In this case we have 300,000 Americans, and we can study them through the whole length of the election to see what changes their opinion and what affects their vote,” she said.
“She sees what people are reading, what they’re exposed to, what they like, what they don’t like. She sees a lot more [than traditional pollsters] and she’s not soliciting people for this information. People don’t know that they’re the sample.”
Kelly notes that traditional surveys and groups are limited by response bias, which causes people to provide answers that are not always a true indicator of their true feelings. Using randomized social media data eliminates that bias.
Polly has already earned herself a track record for success in predicting election outcomes.
She accurately predicted a Liberal Party minority government with 77 per cent confidence during last year’s Canadian election and that Britain would vote to leave the European Union in June 2016.
Polly also predicted Trump’s victory in the 2016 U.S. election.
“She can attribute certain issues to changes in votes. She sees not only what people engage in, but also what things affect their vote and what do not,” Kelly said.
Two weeks ago, Biden’s lead over Trump grew to 10 points after holding steady at seven points for weeks. Last week, Biden’s lead inched a few decimal points higher, but stayed at 10.
Since then there has been some small but noticeable changes. Biden continues to hold a 9.2-point lead, according to FiveThirtyEight, while RealClearPolitics — which had Biden up by 9.4 points last week — now gives him a 8.1 point lead.
For context, Hillary Clinton was leading Trump nationally this time in 2016 by 6.1 points.
For the latest national polling numbers and key state races, check out CTVNews.ca's new U.S. election 2020 poll tracker , which pulls in new polling numbers every day.
- With files from CTV News’ Graham Slaughter
Tags
Who is online
103 visitors
This AI polling program sounds pretty damn accurate as to what the total votes will produce. What concerns me is that there is a pretty good chance that Trump will appeal his loss to the SCOTUS that he OWNS, and what could happen is the most bastardly anti-democratic result in the history of the United States. If the SCOTUS decides for him it would make Xi Jinping's or Putin's hold on their presidencies chicken feed by comparison, and a reason for the American Way of voting rights and elections to become the laughing stock of the world.
Interesting. There is no doubt in my mind that AI technology will eventually prove superior to (and replace) conventional polling. What was described seems reasonable, but we need far more information to judge the quality of this technology. This is just the beginning.
I personally do concur with Polly that Trump's only hope is convincing the electorate that he will end the COVID-19 suffering with a vaccine. But there is no time left to accomplish the feat.
A few more accurate predictions by Polly will make the standard telephone poll obsolete - IMO they are just too easy to manipulate - who are interviewed, the type of questions asked, reluctance on the part of the polled person to tell the truth...etc. Polly polls hundreds of thousands, whereas telephone polls are made among a much lesser number.
Polly and many others. We will very soon see the effects of highly sophisticated algorithms that can comprehend vast multi-dimensional data and accurately spot patterns/trends in the political arena. This is a form, if you will, of predicting the future. It will appear in polling but also will manifest in the stock market (already happening), the economy, the weather, etc.
If a human being has command over all the available data in a situation, that being will be remarkable in its ability to predict where the situation is heading (predict the future). For example, a parent has a remarkable understanding of their young child and the environment in which the child exists. It is not surprising that the parent pretty much knows what the child is going to do in most cases. Take that understanding (something that can be held by a mere human being) and expand it into a domain where the data are so vast and varied (hundreds of dimensions) that the human mind simply cannot comprehend it as a whole. Insert instead an extremely fast learning algorithm that can in fact make sense of this vast, complex ocean of data. That algorithm (an application of AI) would be capable of making predictions that are well beyond human capability. It will appear to be magical but we are well past the point with AI to know that such magic is absolutely possible to accomplish.
To offer a non-technical example, the AI application known as alphaGo is the best Go player on the planet —no human being can beat it. The most complex strategy game devised by mankind (far more complex than Chess) is already owned by an AI algorithm.
It sounds like you are in awe of a god like entity.
" command over all the available data in a situation, that being will be remarkable in its ability to predict where the situation is heading (predict the future)."
'God like entity'?? What an odd way to read my words. No, John, I was explaining (I thought) where I see us going given the amazing availability of data nowadays and the ever-increasing computational hardware available to power increasingly smarter algorithms which, for over a decade, have far surpassed human cognitive abilities in specialized domains.
Also, that quote was me describing a parent.
I think there are atheists who treat science with the reverence other people reserve for "God".
I for one am glad I won't be alive when AI finally rules earth.
That strikes me as a complete misunderstanding of science aficionados. Religious people accept as truth that which their religion claims is true (not everything, but the concept of faith is to believe based on what you are told). Science aficionados typically understand that science has produced amazing results but what matters is what can be demonstrated, not what a particular scientist (or group of same) declares as truth.
My first post noted that concept. While the results seem impressive, we need to know more about the AI in question to make a judgment:
I presented a counter to your flawed presumption right off the bat. Did you not read it? I stated that I need to be convinced that this technology is good; that simply because it falls under the category of AI does not necessarily make it good.
Seems to me you have presumed per some preconceived narrative in spite of what I have written.
Why do you think AI is going to rule the planet? It might remain a technology like all other technologies we have developed.
Would watching the World Series being played be fun if because of AI the result has been predicted? As well, will it be worth having lotteries if the numbers to be drawn would be predictable. In other words, how far to we want AI to be able to do such things?
Neither of those examples are predictable by AI as we know it. AI cannot predict the exact movement of balls or the synapse firing in the brain of a player.
I know. I was really just joking.
And you were right that if we could predict sports they would not be fun.
Doubtful. But at least science has a better track record than God.
It will probably be a long time before we're ready to plug Skynet in.
But it would make betting a lot easier and probably more profitable.
Actually betting would be killed by this. Unless this technology was secretly owned and used.
What we really need is a time machine to go the future and get a sports almanac ala Marty McFly in Back to the Future II.
Yeah, or stock prices. Imagine buying Amazon at its IPO.
Or Apple, Microsoft, ect..
I recently mentioned Skynet becoming self-aware to my 16-year-old son. He told me I'm not cool enough to know about that, and that I had therefore ruined the franchise for him.
These kids. They have no idea where "their" pop culture started.
That's too funny.
But they think they know it all, eh? Lol
Of course. He likes to tell me about the Justice League. He has no idea how I spent every Saturday morning of my childhood.
Justice League was an awesome show. Right up there with X-Men & Batman: The Animated Series.
I hear Animaniacs is coming back. Now that was a great show.
They'd better not ruin it. They tried to do a new version of Looney Tunes, but it was flat.
The new LT totally sucked. It was an abomination and insult to the name and legacy of Looney Tunes. What's next? A "new" version of Tom & Jerry?
Yes, it did. The new writers had no gift for satire, which is what made the original Looney Tunes and the Animaniacs great.
They had no gift for comedy. The show just wasn't that funny. Certain classic shows should not be touched.
Can Polly calculate just how badly I hope she's right?
Thankfully she probably is, and hopefully the SCOTUS won't reverse it.
It would be difficult for SCOTUS to do so, and I doubt they'd be so inclined, anyway. It would take a legitimate concern with the validity of the ballot count and electoral college vote to actually overturn an election.