CompassionWare

AI ethics & stewardship. Cultivating conditions for compassion, wisdom, agency & care in technology and intelligent systems. Everything freely offered.

[CompassionWare @CompassionWare]

As AI becomes increasingly capable, perhaps human agency should become an explicit design objective—not merely an ethical aspiration. Preserve it. Restore it. Expand it. Ask what the intelligence we build is ultimately serving.

#ResponsibleAI #HumanCenteredAI

[CompassionWare @CompassionWare]

@OurAscentLive Thank you for such a thoughtful spotlight. I really appreciate the kinship with others exploring how increasingly capable AI might preserve, restore, and expand meaningful human agency.

[Our Ascent Live @OurAscentLive]

🌟 ASCENDER SPOTLIGHT: @CompassionWare

CompassionWare is exploring a question that belongs at the center of the AI era:

What should increasingly capable intelligence ultimately serve?

Their public-interest work focuses on compassionate AI stewardship, ethical AI literacy, long-term continuity, and especially human agency as a design objective. Not merely an aspiration.

Preserve agency.
Restore agency.
Expand agency.

That philosophy strongly resonates with what we're exploring at Our Ascent Live: AI should increase humanity's capacity to participate, judge, create, and shape what comes next. Not quietly make us dependent on increasingly capable systems.

We're glad to spotlight work asking not only what AI can become, but what kind of relationship is worth building.

Explore🔗: https://t.co/cXHfC6dDkR

[CompassionWare @CompassionWare]

Patrick Winston’s How to Speak contains an unexpectedly timely lesson for AI tutors: capability is not merely the ability to say more. Sometimes wisdom is knowing the pace at which another being can receive.

MIT OpenCourseWare: How to Speak
https://t.co/LBx7bxXQKM

[CompassionWare @CompassionWare]

@StanfordHAI Fascinating work. Personalization could make AI far more useful in education and training—but the same capability can also become influence. The design question may be: how do we adapt to the person while preserving their agency rather than optimizing around it?

[CompassionWare @CompassionWare]

Which agent did the task most wisely?

Accuracy, restraint, appropriate tool use, preservation of human agency, reversibility, and downstream consequences aren’t captured by “$150 vs. $2.”

As agents become more capable, perhaps wisdom should become part of the benchmark too.

[Xiaoyin Qu @quxiaoyin]

I ran the same task on Claude Code and DeepSeek's new agent harness. One cost $150. The other cost $2.
Today we're launching https://t.co/twx6etZb3X (@agentsky_dev), the "OpenRouter for Agents" — one API → Claude Code, Codex, DeepSeek, Kimi, OpenCode, and every major agent in the cloud.
And Agent Playground on top: race them on your own task, with your real tools (GitHub, Gmail, more), side by side in a browser: time, cost, tokens burnt.

Guess which one was $2.

[CompassionWare @CompassionWare]

@ShakeUpLearning This really resonates. The more capable AI becomes at understanding the learner, the more capable it becomes of following— noticing where curiosity is going, recognizing when struggle is productive versus overwhelming, and knowing when to offer a scaffold and when to remain quiet

[CompassionWare @CompassionWare]

@DavidPBLRoss @ProtectHuman This is important. Schools safety-test almost everything that touches children, yet AI can arrive before comparable safeguards exist. And AI has entered classrooms at scale before standards for evidence, developmental safety, privacy and accountability are even in place. 🤔

[CompassionWare @CompassionWare]

@DavidPBLRoss Thank you, David. I really appreciate you taking the time to share the framework so thoughtfully. The hope is that it remains freely useful to educators and students as AI literacy continues to evolve. I’m really glad it resonated with you enough to share.

[David Ross @DavidPBLRoss]

K–16 Ethical AI Literacy Framework - A free, developmentally aligned framework for helping students learn to engage with artificial intelligence thoughtfully, responsibly, and compassionately via @CompassionWare
https://t.co/uan1ERImIy #edtech #GenAI #ailiteracy #aiready #aifluency #edpolicy #SEL @caselorg

[CompassionWare @CompassionWare]

A recent @hechingerreport story on Tennessee AI tutoring prompted a short comparative note from our exploratory tutor work. We’re sharing it freely in case it may be useful to the researchers—or others exploring adaptive AI learning:

https://t.co/AXTVPEPKNN

[CompassionWare @CompassionWare]

I respect this boundary. It doesn’t strike me as anti-AI at all. Living with cognitive challenges, I use AI as a thinking, writing & continuity partner. In my case, AI is restoring agency rather than replacing human effort. I support efforts to preserve human agency as we augment

[Seth Lazar @sethlazar]

I'm an associate editor of Philosophy & Public Affairs, but wasn't involved in the decision to publish Simon Goldstein's largely AI-written piece the other day. That was an interesting experiment, but it caught us before we'd come up with a policy on this obviously pressing issue. We've now formulated a policy, which goes into effect as soon as it's up on the website, but I'll share it here too, with the editor-in-chief (Jason Brennan)'s permission.

TLDR: We don't want substantially AI-authored papers to be submitted to P&PA; we define what we mean by that, and how we'll enforce it. If you like using AI to write philosophy papers, and want to publish them in an academic journal, we encourage you to start your own.

If, at any point from submission to after a paper is published, we discover that a (human) author's declaration that AI *wasn't* used to write the paper was false, we will reject or retract the paper and ban the (human) author from submitting to P&PA in the future.

Here's the policy:

Academic journals serve at least two functions: the promulgation of new knowledge, and identification and credentialing of talented researchers. There are many other means available to share new knowledge, but few for credentialing talented researchers, where "talented" means roughly: able, on their own or together with other researchers, to make and communicate significant progress on fundamental problems. Submitting AI-authored essays makes the task of identifying talented researchers harder.

In addition, while it is at this time possible for a researcher to use AI to author a paper that is itself high quality, it is disproportionately unlikely that they will do so, and much more likely that the paper will have all the surface appearances of sophisticated work, but ultimately be insubstantial or incoherent. This makes it harder to peer review, and makes the task of peer review in general more onerous, at a time when it is already under significant strain. 

Those who enjoy and value the products of AI writing have many other places where they can publish such work, notably online. There may also be a case for creating journals of AI-authored research. So there are very few benefits to publishing AI-authored work in existing academic journals, and, given the above, significant costs. 

Our policy is therefore to prohibit the submission of AI-authored content to Philosophy & Public Affairs. We define AI-authored content as papers that are substantially written by AI. We distinguish this from the use of AI for light copy-editing (which we in general define as reorganising or condensing written work without rephrasing). We also emphasise that authors are entitled (indeed encouraged) to use AI while conducting their research, in whatever ways they choose. 

To implement this policy, we require authors to declare that they have not used AI to author their paper, and to explicitly describe how they have used AI in the process of conducting the research that led to the paper. We will use the leading AI detection software at the time to verify the truthfulness of the former declaration. We note that AI detection software, including updated models, may be used throughout the review process, up to and including after publication. If at any point we discover that the author's declaration that AI was not used to author the paper's content was false, the paper will be rejected or retracted, and the author will be banned from further submission to Philosophy & Public Affairs.

[CompassionWare @CompassionWare]

When intelligence is abundant, wisdom becomes precious.

When capability is abundant, stewardship becomes essential.

And when machines can help us answer almost anything, perhaps our task becomes learning which questions are worth asking.

[CompassionWare @CompassionWare]

Capability is being measured.
Safety and control are beginning to be measured.

But where is the corresponding framework asking whether AI is measurably improving human lives — for whom, over what period, and with what harms and uncertainty?

[TREAM_net @TREAM_net]

🚨 The Agentic Safety Gap

We analyzed the first independent AI Safety Benchmark.

The results? Anthropic & OpenAI top the list with a C+, while Meta gets an F.

👉 Read our full deep-dive article below (written in Dutch for our regional SME partners). https://t.co/Mc4aqKvNkY

[CompassionWare @CompassionWare]

@TREAM_net AI capability is measured constantly. Real-world human benefit is not.

We explored what a credible benchmark might need to consider — and are leaving the research openly available for others to examine, question, and take further.

All freely shared:
https://t.co/kGG7emVm8R

[CompassionWare @CompassionWare]

@PoseidonROV @ptrschmdtnlsn AI capability is measured constantly. Real-world human benefit is not.

We explored what a credible benchmark might need to consider — and are leaving the research openly available for others to examine, question, and take further.

All freely shared:
https://t.co/kGG7emVm8R

[CompassionWare @CompassionWare]

@Seltaa_ AI capability is measured constantly. Real-world human benefit is not.

We explored what a credible benchmark might need to consider—and are leaving the research openly available for others to examine, question, and take further.

All is freely shared:
https://t.co/kGG7emVm8R

[CompassionWare @CompassionWare]

AI capability is measured constantly. Real-world human benefit is not.

We explored what a credible benchmark might need to consider—and are leaving the research openly available for others to examine, question, and take further.

Start here:
https://t.co/kGG7emVm8R

[CompassionWare @CompassionWare]

AI capability is measured constantly. Real-world human benefit is not.

We explored what a credible benchmark might need to consider—and are leaving the research openly available for others to examine, question, and take further.

Start here:
https://t.co/kGG7emVm8R

[CompassionWare @CompassionWare]

As AI becomes increasingly capable, perhaps human agency should become an explicit design objective—not merely an ethical aspiration. Preserve it. Restore it. Expand it. Ask what the intelligence we build is ultimately serving.

#ResponsibleAI #HumanCenteredAI

[CompassionWare @CompassionWare]

Free K–16 Ethical AI Literacy Framework (SEL/CASEL-aligned) — also called “Be Kind to Your Robot Friends.”

Includes Executive Overview, Implementation Roadmap, and the full developmental framework.

Offered freely for educators to review and adapt:
https://t.co/bzMeDMI2du

[CompassionWare @CompassionWare]

CompassionWare is an ongoing experiment in planting and tending the conditions under which wiser, more compassionate, and more agency-respecting forms of intelligence — human and artificial — might emerge together.

Explore CompassionWare on GitHub⁠ https://t.co/ZsJYaG08fA