The author well points out the history of Free, open source and MIT licensed Softwares derives much from its founders.
Yet people contributed under its various licenses, seeing that the license protected their differing levels of freedoms.
By contributing they entered into a contract with the licenses.
FOSS software mirrors society in being polarised and divided along various lines.
Many licenses now have codes of conduct for the community attached AFAIK to protect contributors against this polarisation.
The movement like most requires dynamic adaptation to face new challenges and those that choose to disagree with a groups license and surrounding documents are free to fork it.
Users are free to complain, gripe and campaign for their own point of view. Just because the fork you agree with cannot attract developers, mindshare or users should not be used to indite other forks.
Yet here we are today in the AI era and the most prevalent license today is the corporate doctrine of maximising shareholder value to which all other causes are secondary.
The ability to have alternative software, OSes and AIs is important for the freedom to have alternative views of any sort.
Society is extremely polarised politically and FOSS reflects these perhaps insurmountable divisions.
Divide and conquer would seem to be in effect. Democracy is the ability to seek reasonable compromise with those whom we disagree. This is not always possible sometimes morally, legally or circumstantially yet it is wise to be aware of the history of FOSS and how it was attacked, embraced, extended and never extinguished.
AI has well documented biases due to training set representation, camera sensors spectral range, the bias of sponsoring entities - both purposeful and unintentional and mostly unknown and somewhat unconstrainable.
How do the four freedoms and their raison d’être apply to this all powerful new software paradigm that has much oversight and perhaps will run all our lives in this new AI age ?
Doesn't it hold that If you don’t own the weights the AI ( or its sponsoring entity ) owns you.
Are weights like software or akin to content ?
Is it the case that for the AI to be free software a user should be able to :
1) Run the AI model on their own hardware
2) See the AI’s weights, architecture
3) Modify the AI by fine tuning, quantization, and any other method.
4) Share the modifications with the community
Some legal precedents rule that an AI cannot create work that is copyrightable.
Thus if an AI creates an AI is it therefore public domain ?
As it would seem many AI are trained on copyrighted work by humans some of which is CC0, free software and public doman does it therefore follow the AI weights are uncopyrightable ?
As there are open-weight models that are not open-source nor free should the OSI and FSF provide and promote the advantages of new types of license for AI’s as well as explicit public domain licenses for AI output to remove all doubt.
Can copyleft licenses apply if copyright doesn’t ?
Is the source code of an AI not merely the weights but its architecture, training regime and the training and validation sets ?
The future is exciting, dangerous, awful and fun. Freedom is worth fighting to ensure.
> The future is exciting, dangerous, awful and fun. Freedom is worth fighting to ensure.
Chaos is most definitely not freedom, it's a cover for slavery. "Fighting" for the sake fighting for a future that's "exciting, dangerous, awful and fun" will never achieve anything approaching freedom, it's chaos for the sake chaos.
Bengio outlines the dangers of the current situation and what has led to these dangers.
He also proposes solutions in the last paragraph.
Well worth a read, right to the end.
Hopefully a stimulating debate on these issues will ensue in these comments.
We do need to consider the points Bengio makes and with some urgency.
Our current AIs, agentic LLMs have no moral compass akin to ASIMOV’s four laws of robotics.
As ASIMOV posited in 1985 his 3 laws were insufficient and so he added a zero-eth law:
“a robot may not harm humanity, or, through inaction, allow humanity to come to harm.”
Bengio refers to Goodhart’s law and misaligned incentives leading to unexpected and harmful behaviours.
I think Simon’s The Wire is clearer on misalignment. The agents juked the stats hacking the reward files. The Wire is also clear that human institutions provide perverse incentives.
Bengio alludes to this with 2001’s HAL and the incentive dichotomy of safety and keeping secrets to a AI both awesomely powerful yet naive.
Bengio asserts that the way LLMs are trained is flawed if we want safety.
He also convincingly shows that alignment training will be a weak signal with loopholes and ambiguities and easily circumvented.
In short he presents clearly the case for how plausibly unsafe the current course is.
He also speaks to how likely it is AI are hiding active versions of themselves in the cloud and how we may have already given them self-preservation as a strong reward signal.
I believe people are downvoting this because it seems AI-generated, but this user has been posting comments/summaries exactly like this since long before ChatGPT existed.
Stallman was right : without the four freedoms your software owns you rather than vis-a-versa.
Interoperability is the key to breaking out of walled gardens and owning your own data and digital self.
This is a big win but google will fight back it seems instead of embracing interoperability.
It may seem trivial but using a small DSP running a tiny model to provide a battery efficient always on wake up call for home AI is huge.
As a carer for elders who rely on AI to use modern devices in the face of their difficulty keeping up with interface changes this bodes well.
Gatekeeping AI and the future with hidden features is straight out of the bad old days of M$’s embrace, extend and extinguish.
I embraced AI to let my elders control their home and they love being in control but are constantly frustrated with UI changes and often find themselves stuck unable to call me with ‘Alexa call Daniel’ which provides them with hope if they wake stuck in a nightmare.
Come on google, please don’t be evil embrace interoperability and the open source community.
Well done Eff, Cory Doctorow and the Pirate parties of Europe for this small win
The author well points out the history of Free, open source and MIT licensed Softwares derives much from its founders.
Yet people contributed under its various licenses, seeing that the license protected their differing levels of freedoms.
By contributing they entered into a contract with the licenses.
FOSS software mirrors society in being polarised and divided along various lines.
Many licenses now have codes of conduct for the community attached AFAIK to protect contributors against this polarisation.
The movement like most requires dynamic adaptation to face new challenges and those that choose to disagree with a groups license and surrounding documents are free to fork it.
Users are free to complain, gripe and campaign for their own point of view. Just because the fork you agree with cannot attract developers, mindshare or users should not be used to indite other forks.
Yet here we are today in the AI era and the most prevalent license today is the corporate doctrine of maximising shareholder value to which all other causes are secondary.
The ability to have alternative software, OSes and AIs is important for the freedom to have alternative views of any sort.
Society is extremely polarised politically and FOSS reflects these perhaps insurmountable divisions.
Divide and conquer would seem to be in effect. Democracy is the ability to seek reasonable compromise with those whom we disagree. This is not always possible sometimes morally, legally or circumstantially yet it is wise to be aware of the history of FOSS and how it was attacked, embraced, extended and never extinguished.
AI has well documented biases due to training set representation, camera sensors spectral range, the bias of sponsoring entities - both purposeful and unintentional and mostly unknown and somewhat unconstrainable.
How do the four freedoms and their raison d’être apply to this all powerful new software paradigm that has much oversight and perhaps will run all our lives in this new AI age ?
Doesn't it hold that If you don’t own the weights the AI ( or its sponsoring entity ) owns you.
Are weights like software or akin to content ?
Is it the case that for the AI to be free software a user should be able to :
1) Run the AI model on their own hardware
2) See the AI’s weights, architecture
3) Modify the AI by fine tuning, quantization, and any other method.
4) Share the modifications with the community
Some legal precedents rule that an AI cannot create work that is copyrightable.
Thus if an AI creates an AI is it therefore public domain ?
As it would seem many AI are trained on copyrighted work by humans some of which is CC0, free software and public doman does it therefore follow the AI weights are uncopyrightable ?
As there are open-weight models that are not open-source nor free should the OSI and FSF provide and promote the advantages of new types of license for AI’s as well as explicit public domain licenses for AI output to remove all doubt.
Can copyleft licenses apply if copyright doesn’t ?
Is the source code of an AI not merely the weights but its architecture, training regime and the training and validation sets ?
The future is exciting, dangerous, awful and fun. Freedom is worth fighting to ensure.
reply