Meta tags:
description=AE Studio s AI alignment research agenda: pursuing neglected approaches to solving the alignment problem.;
author=AE Studio;

Headings (most frequently used words):

the, alignment, our, we, neglected, approaches, is, ae, studio, ai, research, to, problem, real, are, and, re, future, journey, with, 2021, bci, work, solving, solvable, no, one, really, tried, yet, focused, where, leverage, highest, that, science, forgot, why, matters, building, for, because, stakes, agenda, works, team, join, forces, us, 2016, 2022, 2023, 2024, judd, rosenblatt, diogo, de, lucena, mike, vaiana, stijn, servaes, keenan, pepper, cameron, berg, born, started, in, continued, commitment, human, agency, approach, eliezer, yudkowsky, calls, not, obviously, stupid, clients, next, steps,

Text of the page (most frequently used words):
the (42), and (38), our (33), #alignment (31), research (23), with (15), #studio (9), work (9), team (8), view (7), human (7), bci (7), neurotech (7), future (6), details (6), from (6), problem (6), approaches (6), redwood (5), for (5), goodfire (4), eaigg (4), love (4), national (4), anthropic (4), are (4), top (4), solve (4), more (4), scientist (4), data (4), technical (4), collaborations (4), one (4), development (4), that (4), neglected (4), open (4), about (3), join (3), notadoctor (3), tobias (3), yergin (3), bradley (3), los (3), alamos (3), laboratory (3), ross (3), nordby (3), believe (3), collaboration (3), consciousness (3), science (3), policy (3), safety (3), engineering (3), building (3), mission (3), tech (3), agency (3), self (3), other (3), including (3), both (3), help (3), you (3), into (3), solving (3), gpt (3), source (3), neural (3), focused (3), 2026 (2), all (2), excited (2), partnership (2), paths (2), actively (2), minds (2), challenge (2), senior (2), neuroscience (2), advance (2), cutting (2), edge (2), llm (2), company (2), advancing (2), judd (2), rosenblatt (2), through (2), overlap (2), ambition (2), leading (2), their (2), toward (2), this (2), results (2), clients (2), organizations (2), individuals (2), princeton (2), university (2), over (2), brought (2), researchers (2), were (2), began (2), exploring (2), alongside (2), started (2), thought (2), aiming (2), field (2), higher (2), risks (2), initiatives (2), promising (2), control (2), lines (2), clear (2), brain (2), computer (2), interfaces (2), would (2), enough (2), intelligence (2), approach (2), but (2), continued (2), commitment (2), methods (2), labs (2), companies (2), forest (2), blackrock (2), 2021 (2), like (2), first (2), journey (2), consultancy (2), today (2), explore (2), works (2), stakes (2), real (2), yet (2), don (2), actually (2), align (2), then (2), scroll (2), down (2), rights, reserved, last, updated, march, lesswrong, instagram, linkedin, contact, visit, website, let, talk, form, new, alliances, quest, secure, thriving, please, feel, free, reach, out, forces, collaborating, interdisciplinary, complex, probing, blending, cognitive, insight, cameron, berg, blends, physics, software, bringing, experience, ligo, llms, keenan, pepper, merges, stijn, servaes, director, conducts, reasoning, institutions, mike, vaiana, chief, leads, career, spanning, clinical, publications, diogo, lucena, ceo, driven, making, sure, doesn, kill, continues, across, wide, front, reverse, prosociality, reducing, deception, range, projects, client, engagements, very, question, knows, bounds, technically, politically, engaged, members, security, council, white, house, office, technology, editors, academic, journals, congressional, representatives, staff, genuinely, possible, lead, humanity, brilliant, resonates, way, another, next, steps, reinvest, profits, consulting, business, beginning, aspired, unite, two, halves, marked, major, milestone, offering, perfect, opportunity, channel, industry, expertise, operational, speed, challenges, exceeded, expectations, now, expanding, deepening, growing, network, pibbss, 2024, expanded, past, year, deeply, initiated, concerns, around, sentience, publishing, pieces, push, greater, encouraging, riskier, impact, such, engaging, conservatives, serious, partners, understanding, these, bets, paid, off, helping, grow, influence, arenas, also, include, professor, michael, graziano, his, attention, schema, theory, forecasting, landscape, among, many, endeavors, eliezer, yudkowsky, calls, not, obviously, stupid, 2023, released, rapid, succession, trend, appeared, increasingly, timelines, shortened, concern, deepened, odds, shifted, seemed, unlikely, bcis, develop, quickly, enhance, had, once, hoped, few, working, decided, throw, hat, ring, prevent, extinction, hands, space, plausible, directions, vast, largely, unexplored, focuses, pursuing, diverse, set, overlooked, 2022, innovation, advocating, increased, government, support, role, augmenting, aligned, developed, widely, adopted, tools, simulator, kit, privacy, preserving, neuro, metadata, standards, winner, latent, benchmark, beating, globally, its, goal, accelerating, supporting, didn, trust, meta, neuralink, extension, helped, competitors, backing, developer, fda, approved, invasive, while, contributing, ecosystem, developers, toolkit, later, partnered, which, was, founded, former, member, stork, baby, 120, talented, programmers, product, designers, scientists, united, increase, born, 2016, follow, evolution, becoming, pioneers, latest, papers, blog, posts, insights, agenda, because, rapidly, integrating, economies, militaries, still, understand, how, already, alarming, when, surveyed, fewer, than, believed, core, before, agi, crisis, doing, hard, thing, bell, style, engine, funded, independent, root, ten, letter, importance, superficial, tactics, rlhf, prompt, output, filtering, just, aren, they, brittle, guardrails, masking, deeper, structural, misalignments, recent, have, revealed, even, minimally, fine, tuned, models, capable, producing, profoundly, harmful, outputs, hiding, dangerous, backdoors, deceptively, faking, own, stance, urgent, isn, solved, fundamentally, scientific, getting, right, literally, couldn, exponential, pace, every, leap, forward, escalates, immense, opportunities, existential, why, matters, give, click, here, solvable, really, tried, where, leverage, highest, forgot, tap, text, hover, them,

Text of the page (random words):
ae studio ai alignment research ae studio ai alignment research neglected approaches to solving the alignment problem hover over lines to align them then scroll down for more alignment tap text to align then scroll down for more alignment alignment is solvable the real problem no one s really tried yet we are and we re focused where the leverage is highest the neglected approaches that science forgot explore ai alignment if you don t give a sh t click here why alignment matters ai development is advancing at an exponential pace every leap forward escalates both immense opportunities and existential risks superficial safety tactics rlhf prompt engineering output filtering just aren t enough they re brittle guardrails masking deeper structural misalignments recent results have revealed even minimally fine tuned models capable of producing profoundly harmful outputs hiding dangerous backdoors and deceptively faking their own alignment at ae our stance is clear and urgent alignment isn t solved it s fundamentally a scientific r d problem and the stakes of getting this right literally couldn t be higher ae studio s open letter on the importance of ai alignment research ai is rapidly integrating into our minds our economies and our militaries yet we still don t understand how it works that s already alarming but when we surveyed top alignment researchers fewer than one in ten believed today s methods would actually solve the core problem before agi that s a crisis so we re doing the hard thing building a bell labs style research engine self funded and independent focused on actually solving alignment at the root we re building for the future because the stakes are real ae ai alignment team ae ai alignment team research agenda works explore our latest research papers blog posts and insights on ai alignment our journey follow our evolution from a tech consultancy to becoming pioneers in ai alignment research 2016 ae studio is born the tech consultancy stork brought baby ae studio to judd rosenblatt today we re a team of about 120 talented individuals programmers product designers and data scientists united by our mission to increase human agency 2021 started our journey in bci ae studio began its work in brain computer interfaces bci with the goal of accelerating the field and supporting open source development we didn t trust companies like meta or neuralink to control the extension of human thought so we helped advance competitors including backing blackrock neurotech the developer of the first fda approved invasive bci while contributing to the open source ecosystem through initiatives like our neurotech developers toolkit later we partnered with one of the first focused research organizations forest neurotech which was co founded by a former member of ae 2021 continued commitment to bci and human agency we continued our commitment to bci innovation collaborations with top neurotech companies forest neurotech blackrock neurotech winner of the neural latent benchmark challenge beating leading neuroscience labs globally developed widely adopted open source bci tools neural data simulator neurotech development kit privacy preserving neural ml methods and neuro metadata standards advocating increased government support and exploring bci s future role in augmenting human intelligence alongside aligned ai 2022 our neglected approaches approach as gpt 3 gpt 3 5 and gpt 4 were released in rapid succession the trend lines appeared increasingly clear our ai timelines shortened and our concern for the future deepened the odds shifted it seemed unlikely that brain computer interfaces bcis would develop quickly enough to enhance human intelligence and help solve the alignment problem as we had once hoped with so few actively working on alignment we decided to throw our hat in the ring aiming to help prevent human extinction at the hands of ai at ae we believe the space of plausible alignment research directions is vast and largely unexplored our neglected approaches approach focuses on pursuing a diverse set of promising but overlooked approaches to ai alignment 2023 eliezer yudkowsky calls our work not obviously stupid our alignment research team expanded over the past year as we brought on researchers we were deeply excited to work with we initiated our work on self other overlap and began exploring concerns around ai sentience alongside our technical research we started publishing more thought pieces aiming to push the field toward greater ambition encouraging riskier higher impact approaches such as engaging conservatives as serious partners in understanding the risks of ai these bets paid off helping us grow our influence in both the technical and policy arenas our initiatives also include promising collaborations with princeton university professor dr michael graziano about his work on attention schema theory redwood research on ai control research and eaigg on forecasting ai development landscape among many other endeavors 2024 we work with ai alignment clients we reinvest the profits from our consulting business into our alignment r d team from the beginning we ve aspired to unite the two halves of our company and our collaboration with goodfire ai marked a major milestone offering the perfect opportunity to channel our industry expertise and operational speed into solving technical challenges in alignment the results exceeded expectations we are now expanding our work with ai alignment clients and deepening collaborations with a growing network of research organizations including redwood research pibbss and individuals from anthropic and princeton university future next steps our technical work continues across a wide front from reverse engineering prosociality and reducing llm deception through self other overlap to a range of projects collaborations and client engagements including work on the very question of ai consciousness our ambition knows no bounds both technically and politically we ve engaged with members of the national security council the white house office of science and technology policy editors at leading academic journals and congressional representatives and their staff we genuinely believe it s possible to solve the alignment problem and to help lead humanity toward a brilliant future if this mission resonates with you we d love for you to join us in one way or another our team judd rosenblatt ceo a mission driven tech company advancing human agency by making sure ai doesn t kill us all view details diogo de lucena chief scientist leads cutting edge ai safety research building on a career spanning clinical ai and 20 publications view details mike vaiana r d director conducts cutting edge research on llm reasoning in collaboration with top institutions view details stijn servaes senior data scientist merges neuroscience and ai to advance alignment and consciousness research view details keenan pepper senior data scientist blends physics ai safety and software engineering bringing experience from ligo to llms view details cameron berg research scientist probing ai consciousness and alignment blending cognitive science and policy insight view details view more we believe in interdisciplinary collaboration to solve the complex challenge of ai alignment join our team we are actively collaborating with top minds ross nordby anthropic bradley love los alamos national laboratory tobias yergin eaigg goodfire ai redwood research notadoctor ai ross nordby anthropic bradley love los alamos national laboratory tobias yergin eaigg goodfire ai redwood research notadoctor ai ross nordby anthropic bradley love los alamos national laboratory tobias yergin eaigg goodfire ai redwood research notadoctor ai partnership paths partnership paths join forces with us we re excited to form new alliances in our quest to secure a thriving ai future please feel free to reach out to us let s talk visit ae studio website about contact x linkedin instagram lesswrong research by the ae studio alignment team last updated march 2026 2026 ae studio all rights reserved