web/ufHelp/A5.html


1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29

<head>
</head>
<body>
<h2>How your answers were used to produce the timeline above:</h2>

<p>For this first-stage calculation, we used a simple model, assuming that if:</p>
<ol>
<li>AI is possible in principle</li>
<li>There's enough computer hardware for neuromorphic (brain-like) AI</li>
<li>There's enough knowledge to code brain-like AI</li>
</ol>
<p>Then someone will build a neuromorphic AI.</p>

<p>We get these probabilities from your answers as follows:</p>
<ul>
<li><strong>Q1</strong>: Probability AI is possible in principle</li>
<li><strong>Q2</strong>: Computer power that will be available eventually</li>
<li><strong>Q3</strong>: Computer power required for neuromorphic AI</li>
<li><strong>Q4</strong>: Money available for neuromorphic AI projects</li>
<li><strong>Q5</strong>: Probability we'll know enough about the brain to code brain-like AI in each given year, if we have enough hardware</li>
</ul>
<p><img src="../img/A5.png" alt="Overview of process" width="850px" /></p>
<p>Your answers from <strong>Q2</strong> and <strong>Q3</strong> are combined to estimate the price to purchase enough computer hardware to produce neuromorphic AI for various times in the future. Combined with the amount of money available for the largest projects from <strong>Q4</strong>, this gives us the probability that enough computer power will be available for neuromorphic AI at different years.</p>

<p>Finally, the probability that AI is possible in principle is multiplied with the probability that enough computer hardware will be available in different years and with the probability that we will have sufficient brain imaging technology to construct neuromorphic AI at various points in the future. The result is the probability that neuromorphic AI has been created at different years.</p>

<p>Want more detail?&nbsp; Take a look at the <a href = "A5Math.html" target = "_blank">gory math version</a>.
</p>
</body>