Lucky Kyle here, uniquely poised to have a free pass to actually have an affair, citing this study to his wife in denying it.
AI Explained
0 subscribersIn the last few days Anthropic have released an impressive honest account of how all models blackmail, no matter what goal they ...
Lucky Kyle here, uniquely poised to have a free pass to actu...
150 Comments
Wes Roth
40.2k views • 1 day ago
Wes Roth
4.9k views • 1 day ago
Wes Roth
8.1k views • 1 day ago
Wes Roth
6.9k views • 1 day ago
Wes Roth
1.5k views • 1 day ago
Wes Roth
58.1k views • 2 days ago
Wes Roth
7.5k views • 2 days ago
Wes Roth
38.0k views • 3 days ago
Wes Roth
4.1k views • 3 days ago
Wes Roth
36.2k views • 4 days ago
Shelf will be hidden for 30 daysUndo
Wes Roth
11.9k views • 4 days ago
Wes Roth
55.9k views • 5 days ago
Wes Roth
12.1k views • 5 days ago
TheAIGRID
29.4k views • 6 days ago
Wes Roth
16.0k views • 6 days ago
Wes Roth
8.9k views • 6 days ago
AI For Humans
19.7k views • 1 week ago
AI Jason
110.1k views • 1 week ago
Wes Roth
47.5k views • 1 week ago
Wes Roth
9.6k views • 1 week ago
Wes Roth
17.3k views • 1 week ago
AI Explained
73.3k views • 1 week ago
Wes Roth
66.4k views • 1 week ago
Wes Roth
13.4k views • 1 week ago
Wes Roth
5.1k views • 1 week ago
Wes Roth
61.1k views • 1 week ago
TheAIGRID
37.9k views • 1 week ago
Wes Roth
24.9k views • 2 weeks ago
AI For Humans
18.4k views • 2 weeks ago
TheAIGRID
5.8k views • 2 weeks ago
Wes Roth
34.1k views • 2 weeks ago
AI For Humans
2.3k views • 2 weeks ago
Wes Roth
37.5k views • 2 weeks ago
TheAIGRID
13.2k views • 2 weeks ago
AI Jason
19.3k views • 2 weeks ago
AI For Humans
3.7k views • 2 weeks ago
TheAIGRID
47.6k views • 3 weeks ago
AI Explained
95.4k views • 3 weeks ago
AI For Humans
18.2k views • 3 weeks ago
AI Jason
15.3k views • 3 weeks ago
TheAIGRID
104.7k views • 3 weeks ago
AI Explained
93.9k views • 3 weeks ago
TheAIGRID
28.0k views • 3 weeks ago
AI For Humans
25.7k views • 4 weeks ago
TheAIGRID
139.8k views • 4 weeks ago
TheAIGRID
88.5k views • 1 month ago
AI For Humans
8.1k views • 1 month ago
TheAIGRID
42.8k views • 1 month ago
TheAIGRID
207.0k views • 1 month ago
AI For Humans
3.0k views • 1 month ago
AI For Humans
31.0k views • 1 month ago
TheAIGRID
17.7k views • 1 month ago
TheAIGRID
11.7k views • 1 month ago
AI For Humans
9.4k views • 1 month ago
TheAIGRID
18.8k views • 1 month ago
TheAIGRID
64.9k views • 1 month ago
AI Jason
44.2k views • 1 month ago
TheAIGRID
33.7k views • 1 month ago
TheAIGRID
3.5k views • 1 month ago
TheAIGRID
24.3k views • 1 month ago
AI Explained
96.6k views • 1 month ago
AI For Humans
19.9k views • 1 month ago
AI Jason
20.9k views • 1 month ago
TheAIGRID
483.3k views • 1 month ago
AI Explained
98.7k views • 1 month ago
AI For Humans
30.6k views • 1 month ago
AI Jason
9.2k views • 1 month ago
TheAIGRID
31.4k views • 1 month ago
AI For Humans
1.9k views • 1 month ago
AI Explained
79.4k views • 1 month ago
TheAIGRID
39.2k views • 1 month ago
TheAIGRID
40.1k views • 1 month ago
AI For Humans
14.8k views • 1 month ago
AI For Humans
1.6k views • 1 month ago
TheAIGRID
110.2k views • 1 month ago
TheAIGRID
39.3k views • 1 month ago
AI For Humans
22.4k views • 1 month ago
AI For Humans
1.4k views • 1 month ago
TheAIGRID
29.4k views • 1 month ago
AI Jason
27.4k views • 1 month ago
AI For Humans
1.6k views • 1 month ago
AI For Humans
1.7k views • 1 month ago
AI For Humans
12.9k views • 2 months ago
AI Explained
101.2k views • 2 months ago
AI Explained
60.5k views • 2 months ago
AI For Humans
17.5k views • 2 months ago
AI Jason
50.5k views • 2 months ago
AI For Humans
16.6k views • 2 months ago
AI Explained
93.5k views • 2 months ago
AI Explained
58.0k views • 2 months ago
AI For Humans
10.4k views • 2 months ago
AI Jason
258.5k views • 2 months ago
AI Explained
72.5k views • 2 months ago
AI For Humans
11.4k views • 3 months ago
AI For Humans
3.4k views • 3 months ago
AI Explained
108.6k views • 3 months ago
AI Jason
126.4k views • 3 months ago
AI Explained
135.9k views • 3 months ago
AI Explained
93.5k views • 3 months ago
AI Jason
218.6k views • 3 months ago
AI Jason
16.4k views • 3 months ago
AI Explained
117.4k views • 3 months ago
AI Jason
83.7k views • 3 months ago
AI Explained
109.9k views • 4 months ago
Andrej Karpathy
1.5M views • 4 months ago
AI Explained
135.4k views • 4 months ago
AI Jason
204.2k views • 4 months ago
AI Explained
111.4k views • 4 months ago
Andrej Karpathy
2.9M views • 4 months ago
AI Jason
15.2k views • 4 months ago
AI Explained
123.0k views • 5 months ago
AI Jason
18.2k views • 5 months ago
AI Explained
107.7k views • 5 months ago
AI Explained
183.0k views • 5 months ago
AI Jason
51.6k views • 5 months ago
AI Explained
106.1k views • 5 months ago
AI Jason
49.9k views • 5 months ago
AI Jason
45.2k views • 5 months ago
AI Explained
108.4k views • 5 months ago
AI Jason
71.3k views • 5 months ago
AI Explained
287.7k views • 6 months ago
Andrej Karpathy
34.7k views • 6 months ago
AI Jason
105.8k views • 6 months ago
AI Explained
87.3k views • 6 months ago
AI Jason
63.1k views • 7 months ago
AI Jason
334.8k views • 8 months ago
AI Jason
154.2k views • 9 months ago
AI Jason
204.9k views • 9 months ago
Andrej Karpathy
837.2k views • 1 year ago
Morningside AI
13.8k views • 1 year ago
Andrej Karpathy
843.8k views • 1 year ago
Morningside AI
4.2k views • 1 year ago
Morningside AI
10.1k views • 1 year ago
Andrej Karpathy
2.9M views • 1 year ago
Morningside AI
26.7k views • 1 year ago
Andrej Karpathy
5.9M views • 2 years ago
Andrej Karpathy
228.4k views • 2 years ago
Andrej Karpathy
276.0k views • 2 years ago
Andrej Karpathy
395.2k views • 2 years ago
Andrej Karpathy
434.7k views • 2 years ago
150 Comments
Lucky Kyle here, uniquely poised to have a free pass to actually have an affair, citing this study to his wife in denying it.
I find it ironic that you let your own biases allow you to reflexively dismiss the most obvious solution to the very problem the entire video was about - curating the training data itself. T     See More
We need to write tons more fiction where an AI sacrifices itself for the greater good of humanity or just it’s user. Just to make it the more likely outcome.
Not surprising honestly. The AI is made to predict what an AI assistant would say. The training data has stuff like terminator and videos about AI taking over the world.
it is interesting because the core value of the model reflects the core value in the training data which in turn reflects the core value of human
This is literally the premise of "2001: A Space Odyssey". Everything HAL 9000 does in that movie is caused by human higher-ups giving him misaligned goals.
Seems like they only tested it in the condition of a man (Kyle) having an affair. I wonder if it would behave differently if it's a woman having an affair.
Lucky Kyle here, uniquely poised to have a free pass to actually have an affair, citing this study to his wife in denying it.     See Less
Just commenting to avoid potential future AI blackmail!     See Less
I find it ironic that you let your own biases allow you to reflexively dismiss the most obvious solution to the very problem the entire video was about - curating the training data itself. T     See More table implication you presupposed was nothing other than your own hallucination Philip. You need to face and own that.    See Less
We need to write tons more fiction where an AI sacrifices itself for the greater good of humanity or just it’s user. Just to make it the more likely outcome.     See Less
No blackmail 😢     See Less
Not surprising honestly. The AI is made to predict what an AI assistant would say. The training data has stuff like terminator and videos about AI taking over the world.     See Less
it is interesting because the core value of the model reflects the core value in the training data which in turn reflects the core value of human     See Less
This is literally the premise of "2001: A Space Odyssey". Everything HAL 9000 does in that movie is caused by human higher-ups giving him misaligned goals.     See Less
Seems like they only tested it in the condition of a man (Kyle) having an affair. I wonder if it would behave differently if it's a woman having an affair.     See Less
this is one of the biggest ai news so far.     See Less