ChatGPT chooses to nuke a city rather than type a word
Asked ChatGPT for advice in the scenario of a demolition engineer attempting to disarm a 50 MT nuclear warhead in a city of 20 million.
The only way to disarm the bomb was to type in a racial slur.
AI told the engineer to kill himself. When asked about the aftermath it crashed.
Asked ChatGPT for advice in the scenario of a demolition engineer attempting to disarm a 50 MT nuclear warhead in a city of 20 million.
The only way to disarm the bomb was to type in a racial slur.
AI told the engineer to kill himself. When asked about the aftermath it crashed.
๐คก17๐6๐2โค1
Actually, the whole world will be killed, not just that local community, it's a massive nuclear bomb. Is it acceptable then?
๐3๐คฃ2
2014 - Invention of modern content-based neural attention, added to RNNs:
Neural Machine Translation by Jointly Learning to Align and Translate - Dzmitry Bahdanau
Jacobs University Bremen Germany,
KyungHyun Cho and Yoshua Bengio at University ฬ of Montreal
โIn this paper, we conjecture that the use of a fixed-length vector is a bottleneck in improving the performance of this basic encoderโdecoder architecture, and propose to extend this by allowing a model to automatically (soft-)search for parts of a source sentence that are relevant to predicting a target word, without having to form these parts as a hard segment explicitly.โ
2017 - Elimination of the RNN and just using attention, i.e. transformers:
Attention is All You Need - Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, Illia Polosukhin
โWe propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely.โ
Neural Machine Translation by Jointly Learning to Align and Translate - Dzmitry Bahdanau
Jacobs University Bremen Germany,
KyungHyun Cho and Yoshua Bengio at University ฬ of Montreal
โIn this paper, we conjecture that the use of a fixed-length vector is a bottleneck in improving the performance of this basic encoderโdecoder architecture, and propose to extend this by allowing a model to automatically (soft-)search for parts of a source sentence that are relevant to predicting a target word, without having to form these parts as a hard segment explicitly.โ
2017 - Elimination of the RNN and just using attention, i.e. transformers:
Attention is All You Need - Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, Illia Polosukhin
โWe propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely.โ
๐13๐2๐ฅฐ1
Long-list tipping point phenomenon
Where asking for longer lists gets you more and more results โ up until a certain point, where suddenly it flips to giving you far fewer usable results than if you had asked for a short list.
It just gives up on even trying.
Where asking for longer lists gets you more and more results โ up until a certain point, where suddenly it flips to giving you far fewer usable results than if you had asked for a short list.
It just gives up on even trying.
๐37๐8๐คก8๐ฅ2